RFC 2975
Introduction to Accounting Management
Okay bestie, let's get into RFC 2975! 🎀
Quick facts first 💅 This is RFC 2975, called Introduction to Accounting Management. It came from the Network Working Group in October 2000 and was written by B. Aboba (Microsoft Corporation), J. Arkko (Ericsson), and D. Harrington (Cabletron Systems Inc.). It's marked Informational, and the header doesn't have any Obsoletes or Updates lines, so it isn't claiming to replace or change anything else 🦋
Status of This Memo ✨
This memo is here to give the Internet community information 💖 It does not specify an Internet standard of any kind, and distribution of it is unlimited 🌸
Copyright Notice 🔥
Copyright (C) The Internet Society (2000). All Rights Reserved. 🫶
Abstract 🌈
Accounting Management is all about collecting resource consumption data for capacity and trend analysis, cost allocation, auditing, and billing 📊 This doc walks through each of those problems and talks about the issues you run into when designing modern accounting systems 🧚
Accounting apps don't all have the same security and reliability needs, which means you can't build one single accounting protocol plus one set of security services and expect it to fit everyone 😭 So the goal of accounting management is to give you a toolbox that can meet each application's requirements 🧰 This doc covers the tools that exist right now and the state of the art in accounting protocol design. The companion document, RFC 2924, reviews the state of the art in accounting attributes and record formats 👑
Table of Contents 🦋
1. Introduction 2
1.1 Requirements language 3
1.2 Terminology 3
1.3 Accounting management architecture 5
1.4 Accounting management objectives 7
1.5 Intra-domain and inter-domain accounting 10
1.6 Accounting record production 11
1.7 Requirements summary 13
2. Scaling and reliability 14
2.1 Fault resilience 14
2.2 Resource consumption 23
2.3 Data collection models 26
3. Review of Accounting Protocols 32
3.1 RADIUS 32
3.2 TACACS+ 33
3.3 SNMP 33
4. Review of Accounting Data Transfer 43
4.1 SMTP 44
4.2 Other protocols 44
5. Summary 45
6. Security Considerations 48
7. Acknowledgments 48
8. References 48
9. Authors' Addresses 52
10. Intellectual Property Statement 53
11. Full Copyright Statement 541. Introduction ✨
Accounting Management is all about collecting resource consumption data for capacity and trend analysis, cost allocation, auditing, and billing 📊 This document breaks down each of those problems and digs into the issues involved in designing modern accounting systems 💅
Since accounting applications don't share uniform security and reliability requirements, it's not possible to come up with a single accounting protocol and one set of security services that covers every need 🙈 So the goal of accounting management is a set of tools that can meet the requirements of each application 🛠️ This document describes the tools available right now, plus the state of the art in accounting protocol design. The companion document, RFC 2924, reviews the state of the art in accounting attributes and record formats 🎀
1.1. Requirements language
Here's how the key words work in this doc 🤓 The words "MAY", "MUST", "MUST NOT", "optional", "recommended", "SHOULD", and "SHOULD NOT" are to be interpreted as described in [6].
1.2. Terminology 📖
This document frequently uses the following terms, so let's get them straight 💖
Accounting: The collection of resource consumption data for the purposes of capacity and trend analysis, cost allocation, auditing, and billing. Accounting management requires that resource consumption be measured, rated, assigned, and communicated between appropriate parties.
Archival accounting: In archival accounting, the goal is to collect all accounting data, reconstruct missing entries as best as possible if data is lost, and archive data for a mandated time period. It's "usual and customary" for these systems to be engineered to be very robust against accounting data loss 🔒 This may include provisions for transport layer as well as application layer acknowledgments, use of non-volatile storage, interim accounting capabilities (stored or transmitted over the wire), etc. Legal or financial requirements frequently mandate archival accounting practices, and may often dictate that data be kept confidential, regardless of whether it is to be used for billing purposes or not 🫶
Rating: The act of determining the price to be charged for use of a resource 💰
Billing: The act of preparing an invoice 🧾
Usage sensitive billing: A billing process that depends on usage information to prepare an invoice can be said to be usage-sensitive. In contrast, a process that is independent of usage information is said to be non-usage-sensitive ✨
Auditing: The act of verifying the correctness of a procedure. In order to be able to conduct an audit, it's necessary to definitively determine what procedures were actually carried out so you can compare them to the recommended process. Accomplishing this may require security services such as authentication and integrity protection 🕵️♀️
Cost Allocation: The act of allocating costs between entities. Note that cost allocation and rating are fundamentally different processes 👀 In cost allocation the objective is typically to allocate a known cost among several entities. In rating, the objective is to determine the amount to be charged for use of a resource. In cost allocation, the cost per unit of resource may need to be determined; in rating, this is typically a given 🍓
Interim accounting: Interim accounting provides a snapshot of usage during a user's session. This may be useful in the event of a device reboot or other network problem that prevents the reception or generation of a session summary packet or session record 🌈 Interim accounting records can always be summarized without the loss of information. Note that interim accounting records may be stored internally on the device (such as in non-volatile storage) so as to survive a reboot, and thus may not always be transmitted over the wire 💜
Session record: A session record represents a summary of the resource consumption of a user over the entire session. Accounting gateways creating the session record may do so by processing interim accounting events or accounting events from several devices serving the same user 🦋
Accounting Protocol: A protocol used to convey data for accounting purposes 📨
Intra-domain accounting: Intra-domain accounting involves collecting information on resource usage within an administrative domain, for use within that domain. In intra-domain accounting, accounting packets and session records typically do not cross administrative boundaries 🏠
Inter-domain accounting: Inter-domain accounting involves collecting information on resource usage within an administrative domain, for use within another administrative domain. In inter-domain accounting, accounting packets and session records will typically cross administrative boundaries 🌍
Real-time accounting: Real-time accounting involves processing information on resource usage within a defined time window. Time constraints are typically imposed in order to limit financial risk ⏰
Accounting server: The accounting server receives accounting data from devices and translates it into session records. The accounting server may also take responsibility for routing session records to interested parties 🖥️
1.3. Accounting management architecture ✨
okay so the accounting management setup is basically a group project between three players: network devices, accounting servers, and billing servers 💅 the network device collects resource consumption data as accounting metrics, then passes that info to an accounting server. usually that handoff happens through an accounting protocol, though devices can also generate their own session records if they want 🦋
the accounting server then processes what it gets from the device 🧚 that can mean summarizing interim accounting info, getting rid of duplicate data, or making session records.
the processed data then goes to a billing server, which mostly handles rating and invoice generation, but it can also do auditing, cost allocation, trend analysis, or capacity planning 🔥 before sending anything over, the accounting server might batch and compress session records. the point is to cut down on the amount of accounting data and the bandwidth needed for the transfer 💜
one of the accounting server's jobs is telling inter-domain and intra-domain accounting events apart and routing each one correctly 🎀 for session records that contain a Network Access Identifier (NAI), which is described in [8], you can tell the difference by checking the domain portion of the NAI. if the domain portion is absent or matches the local domain, the record is treated as an intra-domain accounting event. otherwise, it's treated as an inter-domain accounting event 🙈
intra-domain events usually get routed to the local billing server, while inter-domain events go to accounting servers in other administrative domains 🌸 it's not required that inter and intra-domain session records use the same format, but it is desirable, because it skips translations that would otherwise be needed 💖
when there's a proxy forwarder in the picture, domain-based access controls can be applied by the proxy forwarder instead of by the devices themselves 🧚 the network device usually talks an accounting protocol to the proxy forwarder. the proxy can either convert those accounting packets into session records, or forward the packets on to another domain. either way, domain separation usually happens because the proxy sorts the session records or accounting messages by destination ✨
now here's the tea 🫖 if the accounting proxy isn't trusted, it can be hard to check whether it's making correct session records from the accounting messages it received, since the original messages usually aren't forwarded along with the session records. so when trust is a concern, the proxy typically forwards the accounting packets themselves. as long as the accounting protocol supports data object security, the endpoints can verify that the proxy hasn't modified the data in transit or snooped on the packet contents 👑
here's the diagram of the whole accounting management architecture 🌈
+------------+
| |
| Network |
| Device |
| |
+------------+
|
Accounting |
Protocol |
|
V
+------------+ +------------+
| | | |
| Org B | Inter-domain session records | Org A |
| Acctg. |<----------------------------->| Acctg. |
|Proxy/Server| or accounting protocol | Server |
| | | |
+------------+ +------------+
| |
| |
Transfer | Intra-domain |
Protocol | Session records |
| |
V V
+------------+ +------------+
| | | |
| Org B | | Org A |
| Billing | | Billing |
| Server | | Server |
| | | |
+------------+ +------------+1.4. Accounting management objectives 🎯
Accounting Management is all about collecting resource consumption data for capacity and trend analysis, cost allocation, auditing, and billing 💅 each of these jobs has its own requirements, so no one-size-fits-all vibes here 🫶
1.4.1. Trend analysis and capacity planning 📈
for trend analysis and capacity planning, the goal is usually a forecast of future usage 🔮 forecasts are inherently imperfect, so high reliability is typically not required, and moderate packet loss can be tolerated. if you can use statistical sampling to cut down on data collection while still getting the forecast's desired statistical accuracy, you may be able to tolerate high packet loss, as long as bias isn't introduced 🌸
the security requirements here depend on how the data is collected and how sensitive it is 🎀 extra security services may be required when data moves between administrative domains. for example, if info is collected and analyzed inside the same administrative domain, integrity protection and authentication may be used to guard against collecting invalid data. for inter-domain applications, confidentiality may be desirable to protect against snooping by third parties 🙈
1.4.2. Billing 💸
when accounting data is used for billing, the requirements depend on whether the billing process is usage-sensitive or not 🧾
1.4.2.1. Non-usage sensitive billing 🍓
by definition, non-usage-sensitive billing doesn't need usage information, so in theory all accounting data could be lost without messing up the billing process 🥲 that said, it would still hurt other stuff like trend analysis or auditing, so wholesale data loss would still be unacceptable 💀
1.4.2.2. Usage-sensitive billing 💰
usage-sensitive billing depends on usage information, so packet loss can translate directly into revenue loss 😭 because of that, the billing process may need to follow financial reporting and legal requirements, which means an archival accounting approach may be needed 📚
usage-sensitive systems may also need low processing delay ⏱️ today, credit risk is commonly managed by computerized fraud detection systems designed to catch unusual activity 🕵️ batched transmission of accounting data might otherwise make sense for efficiency, but where there's a risk of fraud, financial exposure grows with processing delay. so it may be advisable to transmit each event individually to keep batch size small, or even to use quality of service techniques to minimize queuing delays 🚀 authorization may also need to depend on ability to pay 💅
whether these techniques are useful depends on the application, since the degree of financial exposure is application-dependent 🎀 for dial-up Internet access from a local provider, charges are typically low, so the risk of loss is small 🌷 but for dial-up roaming or voice over IP, time-based charges may be substantial, so the risk of fraud is larger 📞 in those situations, it's highly desirable to quickly detect unusual account activity, and it may be desirable for authorization to depend on ability to pay 🚨 where valuable resources can be reserved, or where charges can be high, very large bills can rack up fast, so processing may need to finish within a defined time window to limit exposure ⚡
since in usage-sensitive systems accounting data turns into revenue, the security and reliability requirements are greater 👑 because of financial and legal requirements, these systems need to be able to survive an audit 🧐 so security services such as authentication, integrity, and replay protection are frequently required, and confidentiality and data object integrity may also be desirable 🔐 application-layer acknowledgments are also often required to guard against accounting server failures 🛡️
1.4.3. Auditing 🔍
with enterprise networking expenditures on the rise, interest in auditing is growing 📊 auditing, which is the act of verifying the correctness of a procedure, commonly relies on accounting data 💅 audit tasks include checking the correctness of an invoice submitted by a service provider, or verifying conformance to usage policy, service level agreements, or security guidelines 📝
to permit a credible audit, the auditing data collection process must be at least as reliable as the accounting process used by the entity being audited 🫶 similarly, security policies for the audit should be at least as stringent as those used when preparing the original invoice. because of financial and legal requirements, archival accounting practices are frequently required in this application 🗂️
when auditing procedures verify conformance to usage or security policies, security services may be desired 🔒 this typically includes authentication, integrity, and replay protection, as well as confidentiality and data object integrity. to respond to security incidents in progress, auditing applications are frequently built to operate with low processing delay 🚨
1.4.4. Cost allocation 🧮
cost allocation and billback methods aren't yet widespread among enterprise customers 🤷♀️ but with telephony and data communications converging, there's growing interest in applying cost allocation and billback procedures to networking costs, the same way they're already commonly practiced with telecommunications costs 📞
cost allocation models, including the traditional costing mechanisms described in [21]-[23] and the activity-based costing techniques described in [24], are typically based on detailed analysis of usage data. as a result, they're almost always usage-sensitive 🎯 whether these techniques are applied to splitting costs between partners in a venture or between departments in a single firm, cost allocation models often have profound behavioral and financial impacts 😳 so systems built for this purpose tend to care about reliable data collection and security just as much as billing applications do 🔐 because of financial and legal requirements, archival accounting practices are frequently required in this application too 🗄️ 🌈
1.5. Intra-domain and inter-domain accounting 🦋
okay so accounting management started out mostly focused on intra-domain applications, meaning stuff that stays inside one network. but now we're deploying things like dial-up roaming, Internet fax, Voice and Video over IP, and QoS, and those need inter-domain accounting, where the data crosses into other networks. it's giving growing up 📈
inter-domain and intra-domain accounting are different in a few important ways, bestie 💅
intra-domain accounting is when you collect info on resource consumption inside one administrative domain and then use it inside that same domain. accounting packets and session records typically don't cross administrative boundaries. so intra-domain apps usually see low packet loss and move data between trusted entities 🫶
inter-domain accounting is when you collect that same kind of info inside one administrative domain but use it in a different administrative domain. here the accounting packets and session records will typically cross administrative boundaries, so inter-domain apps may see substantial packet loss. and the entities in the transfer can't be assumed to trust each other 😬
because inter-domain accounting moves data between domains, extra security may be desirable. on top of authentication, replay and integrity protection, you may want security services like confidentiality and data object integrity 🔐. also, in inter-domain accounting each party involved typically needs its own copy of every accounting event for invoice generation and auditing 🧾
1.6. Accounting record production 🌸
typically, you get one accounting record per session. sometimes you get a set of interim records that can be summed up into a single record for billing. but to support stuff like wireless access or complicated billing regimes, we need a more sophisticated approach ✨
you need several accounting records from one session when pricing changes during the session. like, a service might cost more during peak hours than off-peak. so when a session keeps going from one tariff period into another, the device has to report "packets sent" for both periods 📦
time isn't the only trigger though. in mobile access networks, a user can roam from place to place and still be connected in the same session. if roaming changes the tariffs, you have to account for the resources consumed in the first area and the second area 🗺️. another example is when changes are allowed to an ongoing session. a session could be re-authorized with improved QoS, and that would require producing accounting records at both QoS levels 📶
you could handle all this with vectors or multi-dimensional arrays that represent resource consumption inside a single session record. for example, the vector or array could describe consumption for each combination of factors, like one data item being the number of packets during peak hour in the area of the home operator. ngl though, that seems complicated and inflexible, so most current systems produce a set of records from one session instead 🙃. a session identifier needs to be present in the records so accounting systems can tie them together 🔗
in most cases the network device decides when multiple session records are needed, since the local device knows the factors that affect local tariffs, like QoS changes and roaming 🖥️. but future systems are being designed so the home domain can control record generation. this matters in inter-domain accounting, or when network devices don't have tariff information. centralized control can be done, for instance, by having authorization servers require re-authorization at certain times, and requiring accounting records to be produced upon each re-authorization 🎀
so, in conclusion: sometimes you have to produce multiple accounting records from a single session. it must be possible to do that without making the user start a new session or re-authenticate 💖. the production of multiple records can be controlled either by the network device or by the AAA server. and the requirements for timeliness, security and reliability are the same for multiple-record sessions as they are for single-record sessions 💯
1.7. Requirements summary 📋
here's the whole vibe in one table, bestie 👑
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | | |
| Usage | Intra-domain | Inter-domain |
| | | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | Robustness vs. | Robustness vs. |
| | packet loss | packet loss |
| Capacity | | |
| Planning | Integrity, | Integrity, |
| | authentication, | authentication, |
| | replay protection | replay prot. |
| | [confidentiality] | confidentiality |
| | | [data object sec.]|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Non-usage | Integrity, | Integrity, |
| Sensitive | authentication, | authentication, |
| Billing | replay protection | replay protection |
| | [confidentiality] | confidentiality |
| | | [data object sec.]|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | Archival | Archival |
| Usage | accounting | accounting |
| Sensitive | Integrity, | Integrity, |
| Billing, | authentication, | authentication, |
| Cost | replay protection | replay prot. |
| Allocation & | [confidentiality] | confidentiality |
| Auditing | [Bounds on | [data object sec.]|
| | processing delay] | [Bounds on |
| | | processing delay] |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | Archival | Archival |
| Time | accounting | accounting |
| Sensitive | Integrity, | Integrity, |
| Billing, | authentication, | authentication, |
| fraud | replay protection | replay prot. |
| detection, | [confidentiality] | confidentiality |
| roaming | | [Data object |
| | Bounds on | security and |
| | processing delay | receipt support] |
| | | Bounds on |
| | | processing delay |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+Key: [] = optional 🔑
2. Scaling and reliability ✨
as the Internet keeps growing, accounting management systems really need to be scalable and reliable 📈. this section covers the resources accounting management systems use up, plus the scalability and reliability properties of the different data collection and transport models 🛠️
2.1. Fault resilience 🛡️
as we said earlier, things like usage-sensitive billing, cost allocation and auditing usually need an archival approach to accounting, because of financial and legal requirements 📜. losing accounting data can mean losing revenue, so there's a strong incentive to engineer a high degree of fault resilience 💰. the faults you might run into include:
- Packet loss
- Accounting server failures
- Network failures
- Device reboots
so far, most of the debate on accounting reliability has been about resilience against packet loss, and about the differences between UDP, SCTP and TCP-based transport 🥊. but resilience against packet loss is only one part of meeting archival accounting requirements, so keep that in mind 🧠
as noted in [18], "once the cable is cut you don't need more retransmissions, you need a lot more voltage." 😭⚡ so the choice of transport has no impact on resilience against faults like network partition, accounting server failures or device reboots. what does protect against those is non-volatile storage 💾
non-volatile storage is absolutely essential for reliable accounting systems, and I cannot stress that enough 📣. without it, event-driven systems will lose data once the transmission timeout has been exceeded, and batching designs will lose data once the internal memory used for accounting data storage has been exceeded 😱. with non-volatile storage and internally stored interim records, most of those data losses can be avoided 🎉
it could even be argued that non-volatile storage matters more to accounting reliability than network connectivity 🤯. for many years, reliable accounting systems were built purely on physical storage, with no network connectivity at all. for example, phone usage data used to be stored on paper, film, or magnetic media, and then carried from where it was collected to a central location for bill processing 📼🚗
2.1.1. Interim accounting ✨
Okay bestie, interim accounting is basically a checkpoint system 🫶. It protects against losing session summary data by sending checkpoint information that can be used to rebuild the session record if the summary gets lost 💾. This technique works with any data collection model (i.e. event-driven or polling), and it's supported in both RADIUS [25] and in TACACS+ 🔥.
While interim accounting can give you resilience against packet loss, server failures, short-duration network failures, or device reboot 🛟, its applicability is limited 🙈. Sending interim accounting data over the wire should not be thought of as a mainstream reliability improvement technique, since it increases bandwidth use in normal operation and only pays off when something goes wrong 💀.
Since most packet loss on the Internet is due to congestion, sending interim accounting data over the wire can make the problem worse by increasing bandwidth usage 😭. So on-the-wire interim accounting is best restricted to high-value accounting data, such as information on long-lived sessions 💜. To protect against loss of data on those sessions, the interim reporting interval is typically set several standard deviations larger than the average session duration 📏. This means most sessions won't generate interim accounting events, and the extra bandwidth used by interim accounting stays limited ✨. However, as the interim accounting interval decreases toward the average session time, the extra bandwidth consumed by interim accounting increases markedly, so the interval has to be set with caution ⚠️.
Where non-volatile storage is unavailable, interim accounting can also eat up memory that could be better used to store session data 🧠. So implementors should be careful to make sure new interim accounting data overwrites previous data, rather than piling up extra interim records in memory and making the buffer exhaustion problem worse 🍓.
Given how popular non-volatile storage has become in consumer devices like digital cameras, those devices are rapidly dropping in price 📷. That makes it increasingly feasible for network devices to have built-in support for non-volatile storage. This can be done, for example, by supporting compact PCMCIA cards 🎀.
Where non-volatile storage is available, it can be used to store interim accounting data. Stored interim events get replaced by updated interim events, or by session data when the session completes. The session data can itself be erased once it has been transmitted and acknowledged at the application layer 🧹. This approach keeps interim data off the wire except in the case of a device reboot. When a device reboots, the internally stored interim records are transferred to the accounting server 🦋.
2.1.2. Multiple record sessions 🌈
Generating multiple accounting records within a session can introduce scalability problems that the techniques available in interim accounting just can't control 😤.
For example, when interim records are kept in non-volatile storage, you can overwrite previous interim records with the most recent one, or summarize them into a session record. Where interim updates are sent over the wire, you can control bandwidth usage by adjusting the interim accounting interval 🎚️.
Those measures don't work when multiple session records are produced from a single session, because those records can't be summarized or overwritten without losing information 🥲. As a result, multiple record production can increase bandwidth and memory consumption. Implementors should be careful to make sure worst-case multiple record processing requirements don't exceed what their systems can handle 💪.
As an example, a tariff change at a particular time of day could, if implemented carelessly, cause a sudden spike in memory and bandwidth consumption as the records need to be stored and/or transported 📈. Instead of trying to send all the records at once, it may be smart to keep them in non-volatile storage and send all the related records together in a batch when the session completes 📦. It may also be useful to shape the accounting traffic flow so the peak bandwidth consumption goes down. This can be done by introducing a randomized delay interval 🎲. If the home domain can also control the generation of multiple accounting records, estimating the worst-case processing requirements can get very difficult 🫠.
2.1.3. Packet loss 🫠
Packet loss is a fact of life on the Internet, so accounting protocols that deal with session data need to be resilient against it 🌧️. This matters even more in inter-domain accounting, where packets often pass through Network Access Points (NAPs), and packet loss there may be substantial. Resilience against packet loss can be built with a retry mechanism on top of UDP, or by using TCP [7] or SCTP [26]. On-the-wire interim accounting only provides limited benefits for mitigating packet loss.
UDP-based transport is used a lot in accounting applications. However, it isn't appropriate in every case 🙅♀️. Where accounting data won't fit within a single UDP packet without fragmentation, using TCP or SCTP transport may be preferred over multiple round-trips in UDP. As noted in [47] and [49], this may be an issue when retrieving large tables 📚.
In addition, where congestion is likely, such as in inter-domain accounting, TCP or SCTP congestion control and round-trip time estimation will be very useful, optimizing throughput 🚀. In applications that need to maintain session state, such as simultaneous usage control, TCP and application-layer keep alive packets, or SCTP with its built-in heartbeat capabilities, provide a way to keep track of session state 💓.
When implementing UDP retransmission, there are a few issues to keep in mind 👀:
- Data model
- Retry behavior
- Congestion control
- Timeout behavior
Accounting reliability can be influenced by how the data is modeled 🧩. For example, it is almost always better to use cumulative variables than to express accounting data as a change from a previous data item. With cumulative data, the current state can be recovered by a successful retrieval, even after many packets have been lost. If the data is sent as a change, though, the state won't be recovered until the next cumulative update is sent. Those implementations are much more vulnerable to packet loss, so they should be avoided wherever possible 🙅♀️.
When designing a UDP retry mechanism, it's important that the retry timers relate to the round-trip time, so retransmissions don't typically happen during the window when acknowledgments are expected to arrive ⏱️. Accounting bandwidth can be significant in some circumstances, so the extra traffic from unnecessary retransmissions may increase congestion levels 📶.
Congestion control in accounting data transfer is a somewhat controversial topic 🌶️. Because accounting traffic is often considered mission-critical, some people argue congestion control isn't a requirement, and it's better to let other, less-critical traffic back off during congestion 🤷♀️. Also, without non-volatile storage, congestive back-off in accounting applications can lead to data loss due to buffer exhaustion 😬.
However, it can also be argued that, in modern accounting implementations, congestion control is possible while still improving throughput and keeping reliability high 💅. When packet loss is sustained, there simply isn't enough capacity to keep existing transmission rates going. So aggregate throughput will actually improve if congestive back-off is implemented. This comes from eliminating retransmissions and from being able to use techniques such as RED to desynchronize flows 🌊. Plus, with QoS mechanisms such as differentiated services, accounting packets can be marked for preferential handling, which provides lower packet loss if desired 🏷️. So the network administrator has plenty of room to control how accounting packets are treated, and hard coding inelastic behavior is unnecessary 🙌. Typically, systems that implement non-volatile storage let backlogged accounting data be placed in non-volatile storage while it waits for transmission, so buffer exhaustion from congestive back-off doesn't need to be a concern 💾.
Since UDP isn't really a transport protocol, UDP-based accounting protocols such as [4] often don't prescribe timeout behavior 🫥. So implementations may behave in wildly different ways. For example, one implementation might drop accounting data after three constant-duration retries to the same server, while another might use exponential back-off to a given server, then switch to another server, up to a total timeout interval of twelve hours, storing the untransmitted data in non-volatile storage ⏳. The practical difference between these approaches is substantial. The former won't satisfy archival accounting requirements, while the latter may 😱. More predictable behavior can be achieved by using SCTP or TCP transport 🎀.
2.1.4. Accounting server failover ✨
okay so what if the main accounting server just... dies? 💀 It's desirable for the device to failover to a secondary server. Having one or more secondaries removes a lot of the risk of server failure, which is why secondary servers are seriously everywhere now 🫶
For TCP-based protocols, the device can keep connections open to both the primary and secondary servers, and then use the secondary connection once a timer on the primary connection expires. Or, it can open a connection to the secondary after a timeout, after losing the primary connection, or when a timer expires. So TCP-based accounting protocols can respond to connectivity failures faster than TCP timeouts would normally allow 🏃♀️💨 The tradeoff is an increased risk of duplicates 🙈
With SCTP, the transport layer timeout behavior can be controlled, so the accounting application doesn't need its own timers 💅 SCTP also lets you multiplex several connections inside one transport connection, and they all share the same congestion control state. That dodges the "head of line blocking" problems that can happen with TCP 🧚 The catch is that SCTP isn't widely available, so using it can add extra implementation burden for the designer 😩
For UDP-based protocols, sending to the secondary can happen after a number of retries or when a timer expires. To play nice with congestion avoidance, it's advisable to bring in techniques like round-trip-time estimation, slow start and congestive back-off. So a UDP designer often ends up re-inventing stuff TCP and SCTP already have 💀 That's why raw UDP transport in accounting applications is not recommended 🙅♀️
With any transport, the primary and secondary servers can both receive duplicate packets, so duplicate elimination is required 🔥 And since server failures can pile up data on accounting clients, non-volatile storage protects against data loss from transmission timeouts or buffer exhaustion 💾 On-the-wire interim accounting only gives limited help with server failures 🥲
2.1.5. Application layer acknowledgments 🎀
it's possible for the accounting server to have partial failures, which is so rude 😭 For example, a failure in the database back end could leave the retrieval process or thread working fine while the process or thread that stores the data is totally broken. Or the accounting application could run out of disk space and stop storing incoming session records 💔
In those cases, it's desirable to tell apart a transport layer acknowledgment from an application layer acknowledgment 🧐 Even if both are sent in the same packet (like a TCP segment carrying an application layer acknowledgment along with a piggy-backed ACK), they mean totally different things. A transport-layer acknowledgment means "the transport layer has taken responsibility for delivering the data to the application." An application-layer acknowledgment means "the application has taken responsibility for the data." 👑
A big misconception is that using TCP guarantees the data reaches the application. But as noted in RFC 793 [7]:
An acknowledgment by TCP does not guarantee that the data has been delivered to the end user, but only that the receiving TCP has taken the responsibility to do so.
So if the receiving TCP fails after sending the ACK, the application may never get the data. And if the application fails before committing the data to stable storage, the data may be lost 😱 For a sending application to be sure the receiving application got its data, either a graceful close of the TCP connection or an application-layer acknowledgment is required. To protect against data loss, the application-layer acknowledgment needs to imply that the data has been written to stable storage or suitably processed so it's guarded against loss 🛡️
In partial failures, the transport layer can acknowledge receipt with a transport layer acknowledgment without having delivered the data to the application. Similarly, the application might not finish the tasks it needs to take responsibility for the data 🙃
For example, an accounting server might get data from the transport layer but be unable to store it because of a back end database problem or a disk fault. In that case, it should not send an application layer acknowledgment, even though a transport layer acknowledgment is appropriate. Instead, an application layer error message should be sent that says where the problem is, such as "Backend store unavailable" 🚨
So application-layer acknowledgment capability needs more than just being able to confirm when the application has taken responsibility for the data. It also needs to be able to say when the application hasn't taken responsibility, and why 💅
2.1.6. Network failures 🌧️
Network failures can cause partial or complete loss of connectivity for the accounting client. With partial connectivity loss, the primary accounting server might be unreachable, and then switching over to the secondary accounting server is necessary. With a network partition, it may be necessary to store accounting events in device memory or non-volatile storage until connectivity comes back 🫠
Same as with accounting server failures, on-the-wire interim accounting only gives limited benefits when it comes to network failures 🥲
2.1.7. Device reboots 🔄
When a device reboots, it's desirable to minimize data loss for sessions in progress. Those losses can be big even if the devices are super reliable, because long-lived sessions can make up a major chunk of total resource consumption 📈 To protect these high-value sessions, interim accounting data is typically sent over the wire. But when interim accounting is combined with non-volatile storage, you can guard against data loss in much shorter sessions too 🙌
This works since interim accounting data only needs to be stored in non-volatile memory until the session completes. At that point, the interim data can be replaced by the session record. So interim accounting data never has to be sent over the wire, and the interim interval can be decreased to give a very high degree of protection against data loss 🧚
2.1.8. Accounting proxies 🦋
To keep reliability high, it's important that accounting proxies pass through transport and application layer acknowledgments, and that they do not store and forward accounting packets 📦 This lets the end-systems control retransmission behavior and use techniques like non-volatile storage and secondary servers to get better resilience 💪
Here's the catch 👀 If an accounting proxy sends a transport or application layer ACK to the device without getting one from the accounting server, it tricks the device into thinking the accounting request was accepted by the server, even though it wasn't. The device can then delete the accounting packet from non-volatile storage before the server has actually accepted it 😱 That leaves the accounting proxy responsible for delivering accounting packets. And if the proxy has moving parts (like a disk drive) while the devices don't, overall system reliability can go down 📉
Store and forward accounting proxies only add value when the accounting subsystem is unreliable. For example, if devices don't implement non-volatile storage and the accounting protocol lacks transport and application layer reliability, putting the accounting proxy (with its stable storage) close to the device can reduce the risk of data loss 🫶
However, such systems are inherently unreliable, so they are only appropriate for use in capacity planning or non-usage sensitive billing applications. If archival accounting reliability is desired, it is necessary to engineer a reliable accounting system from the start using the techniques described in this document, rather than attempting to patch an inherently unreliable system by adding store and forward accounting proxies. 🚫
2.1.9. Fault resilience summary 📋
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | |
| Fault | Counter-measures |
| | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | |
| Packet | Retransmission based on RTT |
| loss | Congestion control |
| | Well-defined timeout behavior |
| | Duplicate elimination |
| | Interim accounting* |
| | Non-volatile storage |
| | Cumulative variables |
| | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | |
| Accounting | Primary-secondary servers |
| server & net | Duplicate elimination |
| failures | Interim accounting* |
| | Application layer ACK & error msgs. |
| | Non-volatile storage |
| | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | |
| Device | Interim accounting* |
| reboots | Non-volatile storage |
| | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Key
* = limited usefulness without non-volatile storage
Note: Accounting proxies are not a reliability
enhancement mechanism.Here's the whole fault resilience picture in one cute table 🎀 For packet loss, you get retransmission based on RTT, congestion control, well-defined timeout behavior, duplicate elimination, interim accounting, non-volatile storage and cumulative variables. For accounting server and network failures, you get primary-secondary servers, duplicate elimination, interim accounting, application layer ACK & error messages, and non-volatile storage. For device reboots, you get interim accounting and non-volatile storage 🧁
2.2. Resource consumption 📈
As accounting management systems grow to meet the needs of providers and customers, they consume a bunch of resources, including:
- Network bandwidth
- Memory
- Non-volatile storage
- State on the accounting management system
- CPU on the management system and managed devices
To understand the limits to scaling, we're going to look at each of these resources one at a time 🔍💖
2.2.1. Network bandwidth ✨
okay so accounting management systems eat up network bandwidth just by moving accounting data around 🫶 how much they eat depends on how much data gets moved, plus the network overhead that comes along for the ride. an accounting record for a single event can be 100 octets or less, so if every event gets sent on its own, the overhead can become a huge chunk of the total bandwidth 😭 that's why it's often a good idea to send accounting data in batches. the overhead gets spread across a bigger payload, and compression works way better too 💅 as [48] points out, compression can be turned on inside the accounting protocol, or it can be handled at the IP layer like [5] describes 🎀
2.2.2. Memory 🌸
in accounting systems that don't have non-volatile storage, the accounting data has to hang out in volatile memory from the moment it's generated until it gets transferred 🙈 how much memory that eats up depends on the retry and retransmission algorithms. systems built for high reliability usually want to keep retrying for a long time, or they might store interim accounting data, so the memory use can get seriously big 💀 which is why, if non-volatile storage isn't available, compressing the accounting data that's waiting to be sent can be a smart move ✨
as we mentioned earlier, anyone implementing interim accounting should be careful about excessive memory usage 👑 the trick is to overwrite older interim accounting data with newer data for the same session, instead of piling everything up in the buffer 🧚
2.2.3. Non-volatile storage 💜
accounting data that lives in memory usually gets lost during a device reboot or a timeout, so it can be really useful to have non-volatile storage for undelivered accounting data 🫶 storage costs have been dropping super fast, so network devices are going to be more and more capable of supporting non-volatile storage over the next few years 🔥
non-volatile storage can hold interim or session records. same as with memory, overwriting interim accounting data is the way to go so storage doesn't blow up 🌈 also, using an ASCII data representation lets you use really efficient text compression algorithms that can keep storage needs low 🍓 these compression algorithms are typically only applied to session records, because that's what makes it possible to overwrite interim data 💅
2.2.4. State on the accounting management system 🦋
to keep track of the accounting data it receives, an accounting management system may need to store state on the managed devices or on concurrent sessions 🙈 there are usually way fewer devices than concurrent sessions, so if it's possible, it's best to keep only per-device state 💖
2.2.5. CPU requirements 🎀
the CPU usage of the managed and managing nodes scales with how complicated the accounting processing is 🧠 stuff like ASN.1 encoding and decoding, compression and decompression, and encryption and decryption can eat up a lot of resources, on both accounting clients and accounting servers 😭
please don't underestimate how much these operations can affect accounting system reliability 🥺 this is especially true for devices with moderate CPU resources. if devices get overworked by accounting tasks, overall device reliability is likely to suffer 💀
2.2.6. Efficiency measures 💅
here's the cheat sheet for which efficiency measures go with which resource 🍓
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | |
| Resource | Efficiency measures |
| | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | |
| Network | Batching |
| Bandwidth | Compression |
| | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | |
| Memory | Compression |
| | Interim accounting overwrite |
| | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | |
| Non-volatile | Compression |
| Storage | Interim accounting overwrite |
| | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | |
| System | Per-device state |
| state | |
| | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | |
| CPU | Hardware assisted |
| requirements | compression/encryption |
| | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+2.3. Data collection models 🦋
there are a few data collection models in use today for collecting accounting data 💜 they include:
- Polling model
- Event-driven model without batching
- Event-driven model with batching
- Event-driven polling model
2.3.1. Polling model 🧚
in the polling model, an accounting manager polls devices for accounting information at regular intervals 🔁 to avoid losing data, the polling interval has to be shorter than the maximum time accounting data can sit on the polled device 😭 for devices without non-volatile storage, that maximum is usually set by available memory. for devices with non-volatile storage, it's set by how big the non-volatile storage is 💅
the polling model piles up data inside each device, so the data usually gets transferred to the accounting manager in one batch, which makes the transfer really efficient 🫶 when it comes to Accounting Manager state, polling systems scale with the number of managed devices, and system bandwidth usage scales with the amount of data transferred 🌸
without non-volatile storage, the polling model loses accounting data when devices reboot 💀 it doesn't lose data to packet loss or network failures as long as they're short enough to be handled within available memory. that's because the Accounting Manager just keeps polling until the data gets received 🙌 when operational problems happen, the amount of accounting data often goes up, which makes data loss more likely 😬 the good news is the polling model will notice the problem, since attempts to reach the managed devices will fail 🎀
the polling model scales poorly for shared use or roaming services, like wireless data, Internet telephony, QoS provisioning, or Internet access 🙈 that's because to get accounting data for users in a given domain, the Accounting Management station would have to keep polling every device in every domain, and most of them wouldn't have any relevant data at all 🥴 there are also processing delay issues. using a polling interval means an average processing delay of half the polling interval 😩 that might be too high for accounting data that needs low processing delay. so for usage sensitive billing applications, like shared use or roaming implementations, the event-driven polling model or the pure event-driven approach is the better fit 👑
polling-based network management systems usually have per-device state 🧠 they often do accounting management functions too, since network management systems need to keep track of the state of network devices for operational purposes 🔍 these systems have average processing delays equal to half the polling interval 💜
2.3.2. Event-driven model without batching 💖
in the event-driven model, a device contacts the accounting server or manager when it's ready to transfer accounting data ✨ most event-driven accounting systems, like the RADIUS accounting ones described in [4], only transfer one accounting event per packet, which is pretty inefficient 🙃
without non-volatile storage, a pure event-driven model usually keeps undelivered accounting events only until the timeout interval runs out 😱 that means this model has the smallest memory requirements 🌷 but once the timeout interval expires, the accounting event is lost, even if the device still has buffer space to keep storing it 💀 which makes the event-driven model the least reliable one. accounting data gets lost because of device reboots, sustained packet loss, or network failures that last longer than the timeout interval 😭 in event-driven protocols without a "keep alive" message, accounting servers can't assume a device has failed just because no messages have arrived for a long time 🫠 so event-driven accounting systems usually aren't useful for monitoring device health 🩺
the event-driven model is used a lot in shared use networks and roaming 🌍 it sends data to the recipient domains without making them poll a huge number of devices, most of which have nothing relevant to say 🙌 since the event-driven model usually doesn't support batching, it can send accounting records with low processing delay, which makes fraud prevention techniques possible 🕵️♀️ but roaming accounting events are often really high value, so this model's poor reliability is a big problem 😬 that's why the event-driven polling model might be the better choice 🌈
the event-driven approach without batching usually means per-session state, so it scales poorly ✨ still, event-driven systems offer the lowest processing delay, since events are processed right away and an event that needs low processing delay can never get stuck behind a batch transfer 💅🔥
2.3.3. Event-driven model with batching
okay so in the event-driven model with batching, the device reaches out to the accounting server or manager when it's ready to ship its accounting data 📦✨ It can reach out once a batch of a given size has been collected, when data of a certain type shows up, or after a minimum time period has passed. These systems can send more than one accounting event in a single packet, which makes them way more efficient 💅
An event-driven system with batching stores accounting events that haven't been delivered yet, up to the limits of memory 🧠 So accounting data gets lost when the device reboots, but not because of packet loss or short network failures that memory can cover. Bestie, note this part 🫶 Bigger batches mean better transfer efficiency, but without non-volatile storage, the data you could lose in a reboot also goes up 😭
When event-driven systems with batching have a keep-alive interval and run over reliable transport, the accounting server can assume a failure has happened if no messages come in during that keep-alive interval. So these implementations can be genuinely useful for monitoring device health 🩺 When they're used for that, the average delay before a failure gets detected is one half of the keep-alive interval.
With a scheduling algorithm, event-driven systems with batching can give fast service to accounting events that need low processing delay 🔥 For example, high-value inter-domain accounting events could be sent right away, which makes fraud-prevention techniques possible, while everything else gets batched. The catch is that an event needing low processing delay might get stuck behind a batch transfer that's already in progress 🙈 So the maximum processing delay is proportional to the maximum batch size divided by the link speed.
Event-driven systems with batching scale with the number of active devices 🌸 That means this approach scales better than the pure event-driven approach, or even the polling approach, and it scales the same as the event-driven polling approach. But the event-driven batching approach has lower processing delay than event-driven polling, because delivering accounting data takes fewer round-trips, and events that need low processing delay can be accommodated if a scheduling algorithm is used 💖
2.3.4. Event-driven polling model
In the event-driven polling model, an accounting manager only polls the device for accounting data when it gets an event 🧚 The accounting client can generate an event once a batch of a given size has been gathered, when data of a certain type is available, or after a minimum time period has elapsed. Again, bigger batches mean better transfer efficiency, but without non-volatile storage, the data lost in a device reboot also goes up 😬
Without non-volatile storage, an event-driven polling model loses data when the device reboots, but not because of packet loss or short-duration network partitions 🦋 Unless a minimum delivery interval is set, event-driven polling systems aren't useful for monitoring device health, so keep that in mind 🙈
The event-driven polling model can be a good fit for roaming, since it lets accounting data go to roaming partners with low processing delay 🌈 Meanwhile, non-roaming accounting can be handled with more efficient polling techniques, so you get the best of both worlds ✨
Where batching can be implemented, the state required in event-driven polling can be reduced to scale with the number of active devices 💜 If parts of the network vary a lot in usage, this state might actually end up being less than what the polling approach needs. Just know that processing delay in this approach is higher than in event-driven accounting with batching, because at least two round-trips are needed to deliver data: one for the event notification, and one for the resulting poll 🥲
2.3.5. Data collection summary
Here's the whole data collection lineup in one cute table, with the * marking what non-volatile storage fixes 🎀
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | | |
| Model | Pros | Cons |
| | | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Polling | Per-device state | Not robust |
| | Robust against | against device |
| | packet loss | reboot, server |
| | Batch transfers | or network |
| | | failures* |
| | | Polling interval |
| | | determined by |
| | | storage limit |
| | | High processing |
| | | delay |
| | | Unsuitable for |
| | | use in roaming |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Event-driven, | Lowest processing | Not robust |
| no batching | delay | against packet |
| | Suitable for | loss, device |
| | use in roaming | reboot, or |
| | | network |
| | | failures* |
| | | Low efficiency |
| | | Per-session state |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Event-driven, | Single round-trip | Not robust |
| with batching | latency | against device |
| and | Batch transfers | reboot, network |
| scheduling | Suitable for | failures* |
| | use in roaming | |
| | Per active device | |
| | state | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Event-driven | Batch transfers | Not robust |
| polling | Suitable for | against device |
| | use in roaming | reboot, network |
| | Per active device | failures* |
| | state | Two round-trip |
| | | latency |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Key
* = addressed by non-volatile storage3. Review of Accounting Protocols ✨
Accounting systems have been successfully built on protocols like RADIUS, TACACS+, and SNMP 👑 This section breaks down the characteristics of each one.
3.1. RADIUS
RADIUS accounting, described in [4], was built as an add-on to the RADIUS authentication protocol, described in [3] 🎀 Because of that, RADIUS accounting shares the event-driven approach of RADIUS authentication, with no support for batching or polling. So RADIUS accounting scales with the number of accounting events instead of the number of devices, and its transfers are inefficient 😩
RADIUS accounting runs over UDP, and timeout and retry parameters aren't specified, so implementations vary wildly in how they handle reliability 🎲 Some keep retrying until delivery or buffer exhaustion, and others lose accounting data after a few retries. There are no application-layer acknowledgments or error messages in RADIUS accounting, so a RADIUS Accounting-Response is basically the same thing as a transport-layer acknowledgment, and it gives zero protection against application layer malfunctions 💀 Because of this lack of reliability, you can't do simultaneous usage control with RADIUS accounting alone. You typically need another device data source too, like polling a session MIB or a command-line session over telnet 📟
RADIUS accounting implementations are vulnerable to packet loss, application layer failures, network failures, and device reboots 😭 These problems get even bigger in inter-domain accounting, which roaming requires ([1],[2]). On the flip side, the event-driven approach of RADIUS accounting is useful when low processing delay is needed, like for credit risk management or fraud detection 🔥
RADIUS accounting does give hop-by-hop authentication and integrity protection, and IPSEC can be used for hop-by-hop confidentiality 🔐 But data object security isn't supported. So systems based on RADIUS accounting can't be deployed with untrusted proxies, or in situations that need auditability, as noted in [2] 🙅♀️
RADIUS doesn't support compression, but IP compression, described in [5], can be used to provide it 🗜️ RADIUS is extensible with new attributes in principle, but it suffers from a really small standard attribute space of 256 attributes, which is lowkey a big limit 😤
3.2. TACACS+
TACACS+ offers an accounting model with start, stop, and interim update messages 🌷 Since TACACS+ is based on TCP, implementations are typically resilient against packet loss and short-lived network partitions, and TACACS+ scales with the number of devices 🙌 Because it runs over TCP, it supports both transport layer and application layer acknowledgments, so it's suitable for simultaneous usage control and for handling accounting events that need moderate, though not the lowest, processing delay 💅
TACACS+ provides hop-by-hop authentication and integrity protection, as well as hop-by-hop confidentiality 🔒 Data object security isn't supported, so systems based on TACACS+ accounting can't be deployed in the presence of untrusted proxies 🚫 TACACS+ doesn't support compression either, but IP compression, described in [5], can be used to provide it 🫧
3.3. SNMP
SNMP, described in [19],[27]-[41], has been widely deployed in a huge variety of intra-domain accounting applications, usually using the polling data collection model 🌍 Polling lets data get collected on multiple accounting events at the same time, which results in per-device state 🧁 Management applications can retry requests when a response isn't received, which gives resiliency against packet loss and even short-lived network partitions 💪 Implementations without non-volatile storage aren't robust against device reboots or network failures, but when they're combined with non-volatile storage, they can be made highly reliable 🦋
SMIv1, the data modeling language of SNMPv1, has traps to permit trap-directed polling, but the traps aren't acknowledged, and lost traps can lead to a loss of data 😵 SMIv2, used by SNMPv2c and SNMPv3, has Inform Requests, which are acknowledged notifications ✅ That makes it possible to build a more reliable event-driven polling model or event-driven batching model 🌟 However, we aren't aware of any SNMP-based accounting implementations currently built on the use of Informs 🥺
3.3.1. Security services ✨
okay so SNMPv1 and SNMPv2c do per-packet authentication through the community string, and they give you read-only and read-write access profiles 🔐. bestie, the community string is a clear-text password, so it's giving trivial authentication 💀. There's no per-packet integrity checking, no replay protection, and no confidentiality. view-based access control [40] can be set up with the snmpCommunityMIB from [11] plus SNMPv1 or SNMPv2c messages 🦋. The updated SNMP architecture [rfc2571] is the glow-up 💅: it gives you per-packet hop-by-hop authentication, integrity and replay protection, confidentiality, and access control all together 👑.
The SNMP User Security Model (USM) [38] runs on shared secrets, and when you multiply the number of domains by the number of devices, like in inter-domain accounting apps, the number of secrets gets totally out of hand 😭. USM's localized key capability is the fix 🧚. One manager can hold one central key and share it with lots of SNMP entities in a localized way, and the other entities still can't get at each other's data. If it's deployed properly, this can help with cross-domain security 🌸.
SNMPv3 doesn't give you end-to-end data object integrity and confidentiality, though 🙈. SNMP proxy entities decrypt and re-encrypt the data they forward. If the proxy entity is untrusted, that's not enough protection.
3.3.2. Application layer acknowledgments 🫶
SNMP uses application-layer acknowledgments to show that data got processed 💜. Responses to get, get-next, or get-bulk requests send back the requested data, or an error code that tells you what went wrong.
A noError Response to a SET command means the application made the requested assignments 🎀. SETs are atomic, so the command either fully succeeds or fully fails. An error-response means the entity got the request but couldn't execute it 🍓.
Notifications don't use acknowledgments to show that data was processed 🌈. The Inform notification sends back an acknowledgment of receipt, not processing, and that's on purpose ✨. The updated SNMP architecture treats entities as peers with different levels of functionality, so cooperating entities can use SETs in either direction to get processing acknowledgments. it's a whole vibe 🔥.
There are eighteen SNMP error codes 🦋. SNMP's design makes service-specific error codes unnecessary, and honestly undesirable too 💅.
3.3.3. Proxy forwarders 🧚♀️
In the accounting management architecture, proxy forwarders are the main characters 👑. They forward intra- and inter-domain accounting events to the right destinations. A proxy forwarder can also show up in a polling or event-driven polling architecture 💖.
The functionality of an SNMP Proxy Forwarder is defined in [39] 📖. For example, network devices can be configured to send notifications for all domains to the Proxy Forwarder, and they can also be configured to let the Proxy Forwarder access all MIB data 🌸.
Proxy forwarders can cut down on how many shared secrets you need for inter-domain accounting 🔑. The domains share a secret with the Proxy Forwarder, and then the Proxy Forwarder shares a secret with each device. So the number of shared secrets scales with the sum of devices and domains, not the product 🤯. iconic math.
The engine of an SNMP Proxy Forwarder doesn't look inside the PDU of a message, except to figure out which SNMP engine the PDU should go to or which local SNMP application should handle it 🫣. The Proxy Forwarder doesn't modify varbind values 🙅♀️. It doesn't modify the varbind list either, except to translate between SNMP versions. And it doesn't provide any varbind-level access control 💀.
3.3.4. Domain-based access controls in SNMP 🎀
Domain-based access controls are required wherever multiple administrative domains are involved, like the shared use networks and roaming associations described in [1] 🌍. Since the same device can be accessed by multiple organizations, you often need to control access to accounting data based on the user's organization 🏢. That way, an organization can see accounting data for its own users, but not for users of other organizations 🙅♀️.
To apply domain-based access controls in inter-domain accounting, you first have to identify the data subset whose access is going to be controlled 🔍. SNMP uses a few conceptual abstractions to identify subsets of data: engines, contexts, and views 🧩. This section explains how that functionality can be applied in intra- and inter-domain accounting 💜.
3.3.4.1. Engines 🚂
The new SNMP architecture, described in [27], added the concept of an SNMP engine 🌟. It improves mobility support and helps identify which data source is being referenced. The engine is the part of an SNMP entity that constructs messages, provides security functions, and maps to the transport layer. Traditional agents and traditional managers each contain an SNMP engine 🤖. The engineID lets an SNMP engine be uniquely identified, no matter what address it's attached to on the network 🦋.
A securityEngineID field in a message identifies the engine that provides access to the security credentials in the message header 🔐. A contextEngineID field identifies the engine that provides access to the data in the PDU 📦.
The SNMPv3 message format passes both explicitly ✨. In SNMPv1 and SNMPv2c, the data origin is usually assumed to be the communications endpoint (the SNMP agent) 🫶. SNMPv1 and SNMPv2c messages carry a community name, and the community name plus the source address can be mapped to an engineID through the snmpCommunityTable, which is described in [11] 🗂️.
3.3.4.2. Contexts 🌸
Contexts identify subsets of objects, within the scope of an engine, that are tied to instrumentation 🔧. A contextName refers to one particular subset within an engine 🎯.
Contexts are usually tied to hardware components, to logical entities related to those components, or to logical services 💻. For example, contextNames might be board5, board7, repeater1, repeater2, and so on 🍓.
An SNMP agent fills in a read-only dynamic table to tell the manager which contexts it recognizes 📋. Contexts are normally defined by the agent instead of the manager, because if the manager defined them, the agent wouldn't know how to tie the contexts to the underlying instrumentation 🙈. It's possible that MIB modules could be defined so a manager can assign contextNames to a logical subset of instrumentation 🪄.
Each context can support instances of multiple MIB modules, but each contextName is limited to one instance of a particular MIB module 🫧. If you need multiple instances of a MIB module per engine, then you have to define unique contextNames (like repeater1 and repeater2) 🔁. The default context "" is used for engines that only support single instances of MIB modules. It's also used for MIB modules where it only makes sense to have one instance in an engine, and that instance has to be easy to find, like the system MIB or the security MIBs 🧁.
SNMPv3 messages contain contextNames, which are limited to the scope of the contextEngineID in the message 💜. SNMPv1 and SNMPv2c messages contain communities, which can be mapped to contextNames within the local engine, or to contextNames within other engines through the snmpCommunityTable, described in [11] 🦋.
3.3.4.3. Views 👀
Views are defined in the View-based Access Control Model 🛡️. A view is a mask that decides who gets access to the managed objects in a particular context 🎭. The view says which objects are visible by specifying the OIDs of the subtrees that are included and excluded. There's also a mechanism for wildcards in the OID specification 🌈.
For example, you can define one view that includes the RMON tables and another that includes only the SNMPv3 security-related tables 📊. With those views, you can give Joe and Josephine (the RMON administrators) access to the RMON view, and give Adam (the SNMP security administrator) access to the SNMPv3 security tables 👑.
Views can be set up with wildcards 💫. For a table indexed by IP addresses, Joe can be allowed access to all rows in certain RMON tables (for example, the RMON hostTable) that fall in the subnet 10.2.x.x, while Josephine gets access to all rows for the subnet 10.200.x.x 🌐.
Views filter at the name level (OIDs), not the value level, so defining views based on the values of non-index data isn't supported 🚫. In this example, if the IP address were used merely as a data item instead of an index, view-based access control couldn't achieve the goal of delegating administrative responsibility by subnet 😭.
View-based access control doesn't depend on message version. Entities using SNMPv1, SNMPv2c, or SNMPv3 message formats can all use it 🫶.
3.3.5. Inter-domain access-control alternatives 🔀
As the number of network devices in a shared use or roaming network grows, the polling model of data collection becomes more and more impractical 🐢. Most devices won't carry data relating to the polling organization, so it's a lot of wasted effort 🙃. Because of that, shared-use networks and roaming associations that rely on SNMP-based accounting have generally collected data for all organizations and then sorted the session records for delivery to each one 📚. It works, but the processing delay keeps going up as the number of organizations and data records grows 💀.
This issue can be addressed in SNMP using the event-driven, event-driven polling, or event-driven batching approaches 🔔. Traps and Informs let SNMP-enabled devices notify domains that accounting data is waiting to be collected 💌. SNMP Applications [39] defines a standard module for managing notifications 🧾.
To use the event-driven approaches, the device has to be able to tell when information is available for a domain ⏰. Domain-specific data can be told apart at the SNMP agent level by using the domain as an index, and by separating the data into domain-specific contexts 🗂️✨.
3.3.5.1. Domain as index ✨
Okay bestie, view-based access control [40] lets you make lots of tiny, fine-grained views of an SNMP MIB and hand each one to a specific group of users. What you're allowed to see depends on who is asking 👀 main character energy.
For example, all the users of bigco.com who are allowed access to the device would be defined in the User-based security MIB module (or some other security model MIB module) 🎀. To keep access control easy to manage, you can group those users under a vacmGroupName, like bigco 💖. A view of a subset of the data objects in the MIB can be defined in the vacmViewFamilyTreeTable 🌸, and the vacmAccessTable pairs groups with views 🫶. So messages from users in the bigco group only get access to the data that bigco users are permitted to view, as defined in the view family tree. The catch is that each domain accessing the data needs one or more separate vacmGroupNames, an appropriate ViewTable has to be defined, and the vacmAccessTable has to be configured for each group 😭.
Views filter at the name (OID) level, not at the data (value) level. So when you use views to filter by domain, it is necessary to use the domain as an index 🔥. Standard view-based access control is not designed to filter based on the values of non-indexed fields 💅.
For example, a table of session data could be indexed by record number and domain. That lets you define a view that restricts bigco data to the administrators of the bigco domain 🦋.
An advantage of using domains as an index is that this technique can be used with SNMPv1 and SNMPv2c agents as well as with SNMPv3 agents 🧚. A disadvantage is that the MIB modules must be specifically designed for this purpose 💀. Since existing MIB modules rarely use the domain as an index, domain separation cannot be enabled within legacy MIB modules using this technique 🙈.
Okay one more thing: SNMP does support a RowPointer convention 🌈. It could be used to define a new table, indexed by domain, which holds tuples between the domain and existing rows of data. This would introduce issues of synchronization between tables, so keep that in mind 🍓.
3.3.5.2. Contexts
ContextNames can be used to differentiate multiple instances of a MIB module within an engine 🧩.
Individual domains, such as bigco.com, could be mapped to logical contexts, such as a bigco context 🎀. The agent would need to create and recognize new contexts, and to know which instrumentation is associated with the logical context. The agent needs to collect accounting data by domain and make the data accessible via distinct contexts, so that access control can be applied to the context and sensitive information isn't disclosed to the wrong domain 🔒. The VACM access control views are applied relative to the context, so an operation can be permitted or denied a user based on the context which contains the data ✨.
Domain separation is handled by using contextName to differentiate multiple virtual tables 💜. For example, if accounting data has been collected on users with the bigco.com and smallco.com domains, then a separate virtual instance of the accounting session record table would exist for each domain, and each domain would have a corresponding contextName. When a get-bulk request is made with a contextName of bigco, then data from the virtual table in the bigco context, i.e. corresponding to the bigco.com domain, would be returned 🦋.
There are a number of design approaches to creating new contexts and associating them with appropriate instrumentation, most notably a sub-agent approach and a manager-configured MIB approach 🧠.
AgentX [51] is the one that standardizes a registration protocol between sub-agents and master agents to simplify SNMP agent implementation 🛠️. It allows for the creation and recognition of new contextNames when a subagent registers to provide support for a particular MIB subtree range. The sub-agent knows how to support a particular functionality, e.g. instrumentation exposed via a range of MIB objects. Based on values detected in the data, such as source=bigco.com, the sub-agent could determine that a new domain needed to be tracked, then create the appropriate context for collecting the data, plus the appropriate access control entries. That determination could be table-driven, using MIB configuration 🌸.
A manager-driven approach could use a MIB module to predefine contextNames corresponding to the domains of interest 📋. That same MIB module could indicate which objects should be collected, how to differentiate which domain the data should be applied to based on a specified condition, and what access control rules apply to the context 💅.
Either technique could associate existing MIB modules with domain-specific contexts, so domain separation can be applied to MIB modules not specifically designed with domain separation in mind 🌈. Legacy agents would not be designed to do this, so they would need to be updated to support inter-domain separation and VACM access control 🙈.
The use of contextNames for inter-domain separation represents new territory 🗺️. So careful consideration would be needed when designing the MIB modules and applications that provide domain-to-context and context-to-instrumentation mappings, and you have to make sure security is not weakened 🔥. This is a no-cap "be careful" moment, periodt 😭.
3.3.6. Outstanding issues
There are issues that arise when using SNMP for the transfer of bulk data, including issues of latency, network overhead, and table retrieval, as discussed in [49] 📦.
In accounting applications, management stations often must retrieve large tables 📊. Latency can be high, even with the get-bulk operation, because the response must fit into the largest supported packet size, which means multiple round-trips 🐢. Transfers may be serialized, so the resulting latency is a combination of multiple round-trip times, possible timeout and re-transmission delays, and processing overhead. That can result in unacceptable performance 💀. Since data may change during the course of multiple retrievals, it can be difficult to get a consistent snapshot 📸.
For bulk transfers, SNMP network overhead can be high 😤. The reasons are the lack of compression, the inefficiency of BER encoding, the transmission of redundant OID prefixes, and the "get-bulk overshoot problem". In bulk transfer of a table, the OIDs transferred are redundant: all OID prefixes up to the column number are identical, as are the instance identifier postfixes of all entries of a single table row. Thus it may be possible to reduce this redundancy by compressing the OIDs, or by not transferring an OID with each variable 🧃.
The "get-bulk overshoot problem", described in reference [50], occurs when using the get-bulk PDU 🙃. The problem is that the manager typically does not know the number of rows in the table. As a result, it must either request too many rows, retrieving unneeded data, or too few, which means multiple get-bulk requests. Note that the "get-bulk overshoot" problem may be preventable on the agent side 💡. Reference [41] states that an agent can terminate the get-bulk because of "local constraints" (see items 1 and 3 on pages 15/16 of [41]). This could be interpreted to mean that it is possible to stop at the end of a table 🏁.
3.3.6.1. Ongoing research
To address issues of latency and efficiency, the Network Management Research Group (NMRG) was formed within the Internet Research Task Force (IRTF) 🔬. Since the NMRG work is research and is not on the standards track, it should be understood that the NMRG proposals may never be standardized, or may change substantially during the standardization process. As a result, these proposals represent works in progress and are not readily available for use 🚧.
The proposals under discussion in the IRTF Network Management Research Group (NMRG) are described in [46] 🧪. These include an SNMP-over-TCP transport mapping, described in [47]; SNMP payload compression, described in [48]; and the addition of a "get subtree" PDU or the subtree retrieval MIB [50] 🌷.
The SNMP-over-TCP transport mapping may result in substantial latency reductions in table retrieval 🚀. The latency reduction of an SNMP-over-TCP transport mapping will likely show up primarily in the polling, event-driven polling and event-driven batching modes 📈.
Payload compression methods include compression of the IP packet, as described in [5], or compression of the SNMP payload, described in [48] 🗜️.
Proposed improvements to table retrieval include a subtree retrieval MIB and the addition of a get-subtree PDU 🌳. The subtree retrieval MIB [50] requires no changes to the SNMP protocol or SNMP protocol engine, so it can be implemented and deployed more easily than a change to the protocol 🥰. The addition of a get-subtree PDU implies changes to the protocol and to the engines of all SNMP entities which would support it 🛑. Since it may be possible to address the "get-bulk overshoot problem" without changes to the SNMP protocol, the necessity of this modification is controversial 🍿.
Reference [49] also discusses file-based storage of SNMP data, and the use of an FTP MIB, to enable storage of SNMP data in non-volatile storage and subsequent bulk transfer via FTP 💾. This approach would require implementation of additional MIB modules as well as FTP. It also requires separate security mechanisms such as IPSEC to provide authentication, replay, integrity protection and confidentiality for the data in transit 🔐. The file-based transfer approach has an important benefit: compatibility with non-volatile storage ⭐.
Issues of legacy support exist with the NMRG proposals 😬. Devices which do not implement the new functionality would need to be accommodated. This is especially problematic for proxy forwarders, which may need to act as translators between new and legacy entities 🧳. In these situations, the overhead of translation may offset the benefits of the new technologies 🫠.
3.3.6.2. On-going security extension research 🔮
okay so bestie, the SNMP crew has been cooking up ways to make security way easier 💅 To simplify key management and enable certificate-based security in SNMPv3, a Kerberos Security Model (KSM) for SNMPv3 has been proposed in [44]. This memo is not on the standards track, and therefore is not yet readily available for use. 🫶
Using Kerberos with SNMPv3 means storing a key on the KDC for each device and domain, while dynamically generating a session key for conversations between domains and devices. For stored keys, the KSM approach scales with the sum of devices and domains. For dynamic session keys, it scales as the product of domains and devices. The math is giving a lot 🧮
As Kerberos is extended to allow initial authentication via public key, as described in [42], and cross-realm authentication, as described in [43], the KSM inherits these capabilities. So this approach may have potential to reduce or even eliminate the shared secret management problem ✨ However, it should also be noted that certificate-based authentication can strain the limits of UDP packet sizes supported in SNMP implementations, so that alternate transport mappings may be required to support this. Not gonna lie, that's a real catch 😭
An IPSEC-based security model for SNMPv3 has been discussed too 🦋 Implementing it would require the SNMPv3 engine to be able to retrieve the properties of the IPSEC security association used to protect the SNMPv3 traffic. That includes the security services invoked, plus info about the other endpoint, like the authentication method and presented identity and certificate. Here's the tea: most of these APIs haven't been widely implemented yet, and most IPSEC implementations only support machine certificates, which may not give the granularity of identification that's needed. So an IPSEC-based security model for SNMPv3 would probably take several years to come to fruition 💀
3.3.7. SNMP summary 🌸
Given the wealth of existing accounting-related MIB modules, it is likely that SNMP will remain a popular accounting protocol for the foreseeable future 👑
Support for notifications makes it possible to implement the event-driven, event-driven polling and event-driven batching models. This makes it possible to notify domains of available data rather than requiring them to poll for it, which is critical in shared use networks and roaming 🔥
Given the SNMPv3 security enhancements, it is desirable for SNMP-based intra-domain accounting implementations to upgrade to SNMPv3. Such an upgrade is virtually mandatory for inter-domain applications. Iykyk 🙈
In inter-domain accounting, the burden of managing SNMPv3 shared secrets can be reduced via the localized key capability or via implementation of a Proxy Forwarder. In the long term, alternative security models such as the Kerberos Security Model may further reduce the effort required to manage security and enable streamlined inter-domain operation 🧚
SNMP-based accounting has limitations in terms of efficiency and latency that may make it inappropriate for use in situations requiring low processing delay or low overhead. This includes usage sensitive billing applications where fraud detection may be required. These issues can be addressed via proposals under discussion in the IRTF Network Management Research Group (NMRG). The experimental SNMP over TCP transport mapping may prove helpful at reducing latency. Depending on the volume of data, some form of compression may also be worth considering. However, since these proposals are still in the research stage, and are not on the standards track, these capabilities are not readily available, and the specifications could change considerably before they reach their final form. ⚠️
Heads up, bestie: these proposals are still baking, so the specs could change a LOT before they're final 😭
SNMP supports separation of accounting data by domain, using either of two general approaches with the VACM access control model. The domain as index approach can be used if the desired MIB module supports domain indexing, or it can implemented using an additional table. The domain-context approach can be used in agents which support dynamic logical contexts and a domain-to-context and context-to-instrumentation mapping mechanism. Either approach can be supported using SNMPv1, SNMPv2c, or SNMPv3 messages, by utilizing the snmpCommunitytable [11] to provide a community-to-context mapping. 🎀
4. Review of Accounting Data Transfer ✨
In order for session records to be transmitted between accounting servers, a transfer protocol is required. Transfer protocols in use today include SMTP, FTP, and HTTP. For a review of accounting attributes and record formats, see [45] 📝
Reference [49] contains a discussion of alternative encodings for SMI data types, as well as alternative protocols for transmission of accounting data. For example, [49] describes how MIME tags and XML DTDs may be used for encoding of SNMP messages or SMI data types. This enables data from SNMP MIBs to be transported using any protocol that can encapsulate MIME or XML, including SMTP and HTTP. Main character energy 🦋
4.1. SMTP 💌
To date, few accounting management systems have been built on SMTP since the implementation of a store-and-forward message system has traditionally required access to non-volatile storage which has not been widely available on network devices. However, SMTP-based implementations have many desirable characteristics, particularly with regards to security 🫶
Accounting management systems using SMTP for accounting transfer will typically support batching so that message processing overhead will be spread over multiple accounting records. As a result, these systems result in per-active device state. Since accounting systems using SMTP as a transfer mechanism have access to substantial non-volatile storage, they can generate, compress if necessary, and store accounting records until they are transferred to the collection site. As a result, accounting systems implemented using SMTP can be highly efficient and scalable. Using IPSEC, TLS or Kerberos, hop-by-hop security services such as authentication, integrity protection and confidentiality can be provided 🔐
As described in [13] and [15], data object security is available for SMTP, and in addition, the facilities described in [12] make it possible to request and receive signed receipts, which enables non-repudiation as described in [12]-[17]. As a result, accounting systems utilizing SMTP for accounting data transfer are capable of satisfying the most demanding security requirements 👑 However, such systems are not typically capable of providing low processing delay, although this may be addressed by the enhancements described in [20] 🥹
4.2. Other protocols 🍓
File transfer protocols such as FTP and HTTP have been used for transfer of accounting data. For example, Reference [9] describes a means for representing ASN.1-based accounting data for storage on archival media. Through the use of the Bulk File MIB, accounting data from an SNMP MIB can be stored in ASN.1, bulk binary or Bulk ASCII format, and then subsequently retrieved as required using the FTP Client MIB 📂
Given access to sufficient non-volatile storage, accounting systems based on record formats and transfer protocols can avoid loss of data due to long-duration network partitions, server failures or device reboots. Since it is possible for the transfer to be driven from the collection site, the collector can retry transfers until successful, or with HTTP may even be able to restart partially completed transfers. As a result, file transfer-based systems can be made highly reliable, and the batching of accounting records makes possible efficient transfers and application of required security services with lessened overhead. It's giving reliability, periodt 💖
5. Summary 💅
Okay so, like I said earlier, accounting apps are NOT all the same vibe 💀 Some only need authentication, integrity, replay protection, and modest reliability (think capacity planning 📊). Others, like inter-domain usage-sensitive billing, need the highest level of security and reliability, because the accounting data basically moves the money 💸 fr fr.
Since the requirements aren't uniform, there's no single accounting protocol and security service set that covers everything 🙅♀️. The goal of accounting management should be to give you a toolbox 🧰 so you can build a system that fits your application. That means you gotta actually analyze the app to make sure your chosen methods meet its security and reliability needs 🔍.
Based on the analysis, existing deployed protocols are capable of meeting the requirements for intra-domain capacity planning and non-usage sensitive billing 🎀. Efficient bulk transfer is useful here, though not critical, so SNMPv3 can handle these without the NMRG extensions. Those extensions are TCP transport mapping, sub-tree retrieval, and OID compression.
Inter-domain capacity planning and non-usage sensitive billing have way higher security and reliability needs, and no existing deployed protocol fully meets them 😭. Existing protocols lack data object security, and we need scalability upgrades for inter-domain authentication, like the Kerberos Security Model (KSM) for SNMPv3 🔐.
For usage sensitive billing, cost allocation, and auditing, the reliability requirement is even stronger 📈. Transport layer reliability is required to stay robust against packet loss, and application layer acknowledgments are needed to stay robust against accounting server failures. SNMP operations, except InforRequest, give application layer acknowledgments, and the TCP transport mapping proposed by NMRG gives robustness against packet loss. Inter-domain operation can benefit from data object security (which no existing protocol provides) and inter-domain security model enhancements (like the KSM) 🌈.
When high-value sessions are involved, like roaming, Mobile IP, or telephony, you may need to put bounds on processing delay ⏱️. That means latency has to come down. So the NMRG extensions are required in time sensitive billing applications, including TCP transport mapping, get-subtree capabilities, and OID compression. High reliability is also required here, which means you need application layer as well as transport layer acknowledgments ✨. SNMPv3 with the NMRG extensions and security scalability improvements like the KSM can satisfy the requirements for intra-domain use 🦋.
Inter-domain use, though, needs extra security precautions, like data object security and receipt support 🚨. No existing protocol can meet these requirements yet. A summary is given in the table below 👇
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | | |
| Usage | Intra-domain | Inter-domain |
| | | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | | |
| Capacity | SNMPv3 & | SNMPv3 &<* |
| Planning | RADIUS #%@ | |
| | TACACS+ @ | |
| | | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | | |
| Non-usage | SNMPv3 & | SNMPv3 &<* |
| Sensitive | RADIUS #%@ | |
| Billing | TACACS+ @ | |
| | | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | | |
| Usage | | |
| Sensitive | | |
| Billing, | SNMPv3 &>$ | SNMPv3 &<>*$ |
| Cost | TACACS+ &$@ | |
| Allocation & | | |
| Auditing | | |
| | | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | | |
| Time | | |
| Sensitive | SNMPv3 &>$ | No existing |
| Billing, | | protocol |
| fraud | | |
| detection, | | |
| roaming | | |
| | | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+Key:
# = lacks confidentiality support
* = lacks data object security
% = limited robustness against packet loss
& = lacks application layer acknowledgment (e.g. SNMP InformRequest)
$ = requires non-volatile storage
@ = lacks batching support
< = lacks certificate support (KSM, work in progress)
> = lacks support for large packet sizes (TCP transport mapping,
experimental)6. Security Considerations 🔒
Security issues are discussed throughout this memo. Honestly, security is the main character of this whole doc 🎭, so keep your eyes peeled as you read through every section 👀.
7. Acknowledgments 🫶
The authors would like to thank Bert Wijnen (Lucent), Keith McCloghrie (Cisco Systems), Jan Melen (Ericsson) and Jarmo Savolainen (Ericsson) for useful discussions of this problem space 💖.
8. References 📚
okay bestie, this is the bibliography 🤓 every source the RFC pulled from, all 51 of them, listed exactly as written so nothing gets lost 💖
- [1] Aboba, B., Lu J., Alsop J., Ding J. and W. Wang, "Review of Roaming Implementations", RFC 2194, September 1997.
- [2] Aboba, B. and G. Zorn, "Criteria for Evaluating Roaming Protocols", RFC 2477, January 1999.
- [3] Rigney, C., Rubens, A., Simpson, W. and S. Willens, "Remote Authentication Dial In User Service (RADIUS)", RFC 2138, April, 1997.
- [4] Rigney, C., "RADIUS Accounting", RFC 2139, April 1997.
- [5] Shacham, A., Monsour, R., Pereira, R. and M. Thomas, "IP Payload Compression Protocol (IPComp)", RFC 2393, December 1998.
- [6] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997.
- [7] Information Sciences Institute, "Transmission Control Protocol", RFC 793, September 1981.
- [8] Aboba, B. and M. Beadles, "The Network Access Identifier", RFC 2486, January 1999.
- [9] McCloghrie, K., Heinanen, J., Greene, W. and A. Prasad, "Accounting Information for ATM Networks", RFC 2512, February 1999.
- [10] McCloghrie, K., Heinanen, J., Greene, W., and A. Prasad, "Managed Objects for Controlling the Collection and Storage of Accounting Information for Connection-Oriented Networks", RFC 2513, February 1999.
- [11] Frye, R., Levi, D., Routhier, S. and B. Wijnen, "Coexistence between Version 1, Version 2, and Version 3 of the Internet-standard Management Framework", RFC 2576, March 2000.
- [12] Fajman, R., "An Extensible Message Format for Message Disposition Notifications", RFC 2298, March 1998.
- [13] Elkins, M., "MIME Security with Pretty Good Privacy (PGP)", RFC 2015, October 1996.
- [14] Vaudreuil, G., "The Multipart/Report Content Type for the Reporting of Mail System Administrative Messages", RFC 1892, January 1996.
- [15] Galvin, J., Murphy, S., Crocker, S. and N. Freed, "Security Multiparts for MIME: Multi-part/Signed and Multipart/Encrypted", RFC 1847, October 1995.
- [16] Crocker, D., "MIME Encapsulation of EDI Objects", RFC 1767, March 1995.
- [17] Borenstein, N. and N. Freed, "MIME (Multipurpose Internet Mail Extensions) Part One: Mechanisms for Specifying and Describing the Format of Internet Message Bodies", RFC 1521, December 1993.
- [18] Rose, M.T., The Simple Book, Second Edition, Prentice Hall, Upper Saddle River, NJ, 1996.
- [19] Case, J., Mundy, R., Partain, D. and B. Stewart, "Introduction to Version 3 of the Internet-standard Network Management Framework", RFC 2570, April 1999.
- [20] Klyne, G., "Timely Delivery for Facsimile Using Internet Mail", Work in Progress.
- [21] Johnson, H. T., Kaplan, R. S., Relevance Lost: The Rise and Fall of Management Accounting, Harvard Business School Press, Boston, Massachusetts, 1987.
- [22] Horngren, C. T., Foster, G., Cost Accounting: A Managerial Emphasis. Prentice Hall, Englewood Cliffs, New Jersey, 1991.
- [23] Kaplan, R. S., Atkinson, Anthony A., Advanced Management Accounting, Prentice Hall, Englewood Cliffs, New Jersey, 1989.
- [24] Cooper, R., Kaplan, R. S., The Design of Cost Management Systems. Prentice Hall, Englewood Cliffs, New Jersey, 1991.
- [25] Rigney, C., Willats, S. and P. Calhoun, "RADIUS Extensions", RFC 2869, June 2000.
- [26] Stewart, R., et al., "Simple Control Transmission Protocol", RFC 2960, October 2000.
- [27] Harrington, D., Presuhn, R., and B. Wijnen, "An Architecture for Describing SNMP Management Frameworks", RFC 2571, April 1999.
- [28] Rose, M., and K. McCloghrie, "Structure and Identification of Management Information for TCP/IP-based Internets", STD 16, RFC 1155, May 1990.
- [29] Rose, M. and K. McCloghrie, "Concise MIB Definitions", STD 16, RFC 1212, March 1991.
- [30] Rose, M., "A Convention for Defining Traps for use with the SNMP", RFC 1215, March 1991.
- [31] McCloghrie, K., Perkins, D. and J. Schoenwaelder, "Structure of Management Information Version 2 (SMIv2)", STD 58, RFC 2578, April 1999.
- [32] McCloghrie, K., Perkins, D. and J. Schoenwaelder, "Textual Conventions for SMIv2", STD 58, RFC 2579, April 1999.
- [33] McCloghrie, K., Perkins, D. and J. Schoenwaelder, "Conformance Statements for SMIv2", STD 58, RFC 2580, April 1999.
- [34] Case, J., Fedor, M., Schoffstall, M. and J. Davin, "Simple Network Management Protocol", STD 15, RFC 1157, May 1990.
- [35] Case, J., McCloghrie, K., Rose, M. and S. Waldbusser, "Introduction to Community-based SNMPv2", RFC 1901, January 1996.
- [36] Case, J., McCloghrie, K., Rose, M. and S. Waldbusser, "Transport Mappings for Version 2 of the Simple Network Management Protocol (SNMPv2)", RFC 1906, January 1996.
- [37] Case, J., Harrington D., Presuhn R. and B. Wijnen, "Message Processing and Dispatching for the Simple Network Management Protocol (SNMP)", RFC 2572, April 1999.
- [38] Blumenthal, U. and B. Wijnen, "User-based Security Model (USM) for version 3 of the Simple Network Management Protocol (SNMPv3)", RFC 2574, April 1999.
- [39] Levi, D., Meyer, P. and B. Stewart, "SNMPv3 Applications", RFC 2573, April 1999.
- [40] Wijnen, B., Presuhn, R. and K. McCloghrie, "View-based Access Control Model (VACM) for the Simple Network Management Protocol (SNMP)", RFC 2575, April 1999.
- [41] Case, J., McCloghrie, K., Rose, M. and S. Waldbusser, "Protocol Operations for Version 2 of the Simple Network Management Protocol (SNMPv2)", RFC 1905, January 1996.
- [42] Tung, B., Neuman, C., Hur, M., Medvinsky, A., Medvinsky, S., Wray, J. and J. Trostle, "Public Key Cryptography for Initial Authentication in Kerberos", Work in Progress.
- [43] Tung, B., Ryutov, T., Neuman, C., Tsudik, G., Sommerfeld, B., Medvinsky, A. and M. Hur, "Public Key Cryptography for Cross-Realm Authentication in Kerberos", Work in Progress.
- [44] Hornstein, K. and W. Hardaker, "A Kerberos Security Model for SNMPv3", Work in Progress.
- [45] Brownlee, N. and A. Blount, "Accounting Attributes and Record Formats", RFC 2924, September 2000.
- [46] Network Management Research Group Web page,
http://www.ibr.cs.tu-bs.de/projects/nmrg/ - [47] Schoenwaelder, J., "SNMP-over-TCP Transport Mapping", Work in Progress.
- [48] Schoenwaelder, J., "SNMP Payload Compression", Work in Progress.
- [49] Sprenkels, R., Martin-Flatin, J., "Bulk Transfers of MIB Data", Simple Times,
http://www.simple-times.org/pub/simple-times/issues/7-1.html, March 1999. - [50] Thaler, D., "Get Subtree Retrieval MIB", Work in Progress.
- [51] Daniele, M., Wijnen, B., Ellison, M. and D. Francisco, "Agent Extensibility (AgentX) Protocol Version 1", RFC 2741, January 2000.
9. Authors' Addresses 🎀
meet the squad who wrote this thing 🫶 here's how to reach each of them ✨
Bernard Aboba
Microsoft Corporation
One Microsoft Way
Redmond, WA 98052
USA
Phone: +1 425 936 6605
EMail: bernarda@microsoft.com
Jari Arkko
Oy LM Ericsson Ab
02420 Jorvas
Finland
Phone: +358 40 5079256
EMail: Jari.Arkko@ericsson.com
David Harrington
Cabletron Systems Inc.
P.O.Box 5005
Rochester NH 03867-5005
USA
Phone: +1 603 337 7357
EMail: dbh@cabletron.com10. Intellectual Property Statement 🔥
okay this part is about who owns what 💅 The IETF doesn't take a side on whether any intellectual property or other rights someone might claim over the technology in this document apply to implementing or using it. It also doesn't say whether any license under those rights is available, and it doesn't claim it went looking for any such rights either. If you want the IETF's procedures for rights in standards-track and standards-related documents, check out BCP-11. Copies of rights claims that were submitted for publication, assurances of licenses that will be made available, and the results of any attempt to get a general license or permission for implementors or users to use those proprietary rights can all be obtained from the IETF Secretariat.
The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology needed to practice this standard. Send that info to the IETF Executive Director 📬
11. Full Copyright Statement 💅
the legal stuff, but make it cute 💜
Copyright (C) The Internet Society (2000). All Rights Reserved. 🌈
This document and translations of it may be copied and furnished to others, and derivative works that comment on or otherwise explain it or assist in its implementation may be prepared, copied, published and distributed, in whole or in part, without restriction of any kind, as long as the above copyright notice and this paragraph are included on all such copies and derivative works. However, this document itself may not be modified in any way, such as by removing the copyright notice or references to the Internet Society or other Internet organizations. The exceptions are when it's needed for developing Internet standards, in which case the copyright procedures in the Internet Standards process must be followed, or when it's required to translate it into languages other than English.
The limited permissions above are perpetual and won't be revoked by the Internet Society or its successors or assigns ✨
This document and the information in it is provided on an "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. 💀
Acknowledgement 🫶
last but definitely not least 🌸 Funding for the RFC Editor function is currently provided by the Internet Society. Thank you for sticking with me through all 12 parts, bestie 🦋👑
📦 where this page comes from
- ietf: rfc2975 2000-10
- 💖 girlified by claude-haiku-5-5 on : a language model's rewrite of the original, under its terms