RFC 1679

HPN Working Group Input to the IPng Requirements Solicitation ✨

• Series: Network Working Group 🧚
• RFC number: 1679 🎀
• Category: Informational 📚
• Authors: D. Green, P. Irey, D. Marlow, and K. O'Donoghue, listed with NSWC-DD as their affiliation 🫶
• Date: August 1994 📅
• Obsoletes / Updates: The title block doesn't say it obsoletes or updates anything 💅

---

Status of this Memo

Okay so this memo is just here to provide information for the Internet community 💖. It does not specify an Internet standard of any kind 🙈. Distribution is unlimited, so share it with your whole group chat 🦋.

Abstract

This doc was submitted to the IETF IPng area as a response to RFC 1550 📨. Publishing it doesn't mean the IPng area agreed with any of the ideas inside 😬. If you want to leave comments, send them to big-internet@munnari.oz.au 💌.

Executive Summary

The Navy's High Performance Network (HPN) working group has been looking closely at what mission critical apps need on Navy platforms 🚢✈️. From that study, they pulled out three big buckets of issues for IPng 🪣.

The first bucket is assumptions: current functionality, commercial viability, and transitioning 🫧. The second is general requirements: addressing, an integrated services architecture, mobility, multicast, and rapid route reconfiguration 🔥. The third is additional considerations: fault tolerance, policy based routing, security, and time synchroniztion ⏰.

HPN really wants to team up with the IETF on standards that would apply to mission critical systems 🤝. Their main focus is multicast functionality, an integrated services architecture, and support for high performance subnetworks 💜.

1. Introduction

HPN was set up to study future network architectures for mission critical applications aboard Navy platforms 🛳️. Because of that, they're super invested in how the IPng selection and development process goes 👀. This document came straight out of HPN's own discussions 🗣️.

The whole point of this doc is to lay out what HPN sees as requirements for an IPng protocol set 📋. Honestly, a lot of what they need already exists in current Internet and ISO network protocols ✨. But HPN spotted some needed capabilities that go beyond existing standards 😤.

They sort their topics into three categories 🗂️. The first is assumptions: things HPN believes the IPng process will handle well without specific Navy input 🙌. The second is general requirements: capabilities that existing network protocols don't really cover yet, and that matter a lot to Navy mission critical apps 💅. The third is a set of additional considerations: still important to HPN, but there's no guidance or specific requests they can give yet 🫠.

2. Background

The US Navy set up a program through the Space and Naval Warfare Systems Command called the Next Generation Computer Resources (NGCR) Program 🇺🇸. Its job is to figure out the changing needs for information system technology in Navy mission critical systems 🔍.

The NGCR High Performance Network (HPN) working group was only just created by the NGCR program 🌱. Their mission is to look at high performance networks for future Navy platforms, which means aircraft, surface ships, submarines, and certain shore-based applications ✈️🚤🌊. They're still reviewing Navy needs right now, and the requirements below are based on what HPN understands about those application areas so far 🧠. The time frame for designing, developing, and deploying HPN based systems and subsystems is 1996 into the twenty first century ⏳.

HPN has identified three general problem domains they're targeting within a mission critical environment 🎯. The first is a distributed combat system environment 🖥️. It's kind of like a bunch of workstations running lots of different apps that pull in multiple sources and types of information 🌀. Analog, audio, digital, discrete, graphic, textual, video, and voice information all need to be coordinated so a commander, operator, or any end user gets a single concise view 👑.

The second problem area is the general internetworking environment 🌐. That's about moving information to many heterogeneous systems over various subnetworks 🔀. The third is providing a high speed interconnect for devices like sensors and signal processors ⚡. [1]

2.1. Application Area

The application area of HPN is the communication network that's part of the mission critical systems on Navy platforms 🛰️. The end points, or users, of HPN include humans, computers, and the many devices (cameras, etc.) found on those platforms 📷. What these end points do includes sensor input, signal processing, operator consoles, navigation systems, and so on 🎛️. They're usually grouped into systems both on platforms and at shore-based sites 🏢. These systems handle long range planning, analysis of sensor information, and machinery control in real-time 🕹️.

The information types HPN has identified as required include voice 🎙️, live and pre-recorded audio ranging from voice to CD quality (e.g., from sensors) 🎶, and video at 1 to 30 frames per second in both monochrome and color 🎥. They also need image data, either static or from real-time sensors 🖼️, plus reliable and connectionless data transfer 📦. And then there's very high-bandwidth (gigabits per second) unprocessed sensor data, which is a LOT 🤯.

2.2. Services

Another way to sort the HPN application area is by looking at the user services it has to support 💁‍♀️. Here are some of them:

1. process to process message passing
2. distributed file and database manipulation
3. e-mail (both within the platform and off the platform) 📧
4. teleconferencing (with the platform, between platforms, and across the Internet) 📹
5. video monitoring of various physical environments 👀
6. voice distribution (as a minimum between computer processes and people) 🗣️
7. image services 🖼️
8. time synchronization ⏰
9. name or directory services 📒
10. network and system management 🛠️
11. security services (support of multilevel data security, privacy and protection) 🔐

---

3. Assumptions

These are the assumptions HPN is basically counting on IPng to cover 🤞. HPN thinks they're important enough to call out, even though they aren't requirements HPN is pushing for 🫶.

3.1. Accommodation of Current Functionality

IPng has to support at least the existing functionality 💯. HPN flagged these specific issues:

1. IPng has to provide basic connectionless transfer of information from one end-point to another 📬.
2. IPng has to support multiple subnetwork technologies. That includes, but isn't limited to, Ethernet, FDDI, Asynchronous Transfer Mode (ATM), Fiber Channel, and Scalable Coherent Interface (SCI) 🔌. These are the subnetwork technologies HPN cares about most 💕. Ideally, IPng protocols should be subnetwork independent 🧩.
3. IPng has to support multihomed hosts, meaning a single host supporting multiple different subnetwork technologies 🏠. Multihomed hosts have to be able to steer traffic to the subnetworks they pick 🚦.
4. IPng needs to recognize that it could be just one of several network protocols a host uses 🫧.
5. IPng needs to provide appropriate network management in the finished product 🧑‍💻. Network management is super vital for the apps HPN is focused on 🚨.

3.2. Commercial Viability

Just like in the commercial world, HPN feels really strongly that IPng protocols have to be commercially viable 💼. That includes, but isn't limited to, these issues:

1. IPng protocols have to function correctly. The Navy can't afford network protocol problems in mission critical systems 😰. There needs to be a high degree of confidence that the protocols are technically sound, and that multi-vendor interoperability is achievable 🤝.
2. IPng protocols have to have support from the commercial/industrial community 🏭. HPN says this could first be shown by a strong consensus within the IETF community 🗳️.

3.3. Transition Plan

The Navy has a huge number of existing networks, including Internet and ISO protocols as well as a bunch of proprietary systems 🗄️. At minimum, the IPng effort has to address how to transition from existing IP based networks 🚧. It would also be nice to have some guidance for transitioning from other network protocols, including CLNP and other commonly used ones 💭.

The transition plan for IPng needs to recognize the large existing infrastructure and the lack of funds for a full scale immediate transition 💸. HPN expects there will probably be a long period of co-existence, and that should be addressed too ⏳🫂.

---

4. General Requirements

These are the general requirements HPN considers vital in any network protocol solution 🚨. They're hoping IPng addresses all of them 🤞.

4.1. Addressing

HPN has identified initial addressing requirements 📍. First off, a large number of addresses is needed. The number of addressable entities on a single platform will range from the 100's to 100,000 🤯. The number of large platforms (ships, submarines, shore based sites) will range from a few hundred to several thousand 🚢. On top of that, there will be 500 to 1000 or more small platforms, mostly aircraft ✈️. Since many of these platforms are expected to connect to global networks eventually, the addresses must be globally unique 🌍.

The second requirement is for some form of addressing structure 🧱. This structure should be flexible enough to allow logical structures (not necessarily geographical) to be applied 🗺️. HPN also thinks this matters for implementing efficient routing solutions 🛣️. In addition, the addressing structure has to support multicast group addressing 📣. At a minimum, 2**16 globally unique multicast groups have to be distinguishable per platform 💜.

4.2. Integrated Services Architecture

okay so the HPN working group is hunting for existing and emerging tech that can bring together the services mission critical Navy systems need 🔥 They split the problem into two big classes under the umbrella of integrated services 💅 Class one: it is REQUIRED to support the multiple types of services from section 2.1 in an integrated fashion, so related streams of info can be correlated in time ⏰ bestie, that timing piece is everything 😭

Class two is about predictably managing all the traffic flows tied to those services 🦋 Lots of these services need a PDU delivered inside a specified time window, but mission critical apps are next level picky 💀 In areas with real-time systems, like machinery control, narrower and/or more predictable delivery windows MAY be REQUIRED than for audio or video streams 🎀 The mission critical environment also needs to be able to assign end-to-end importance to instances of communications (i.e., invocations of a particular service). For example, an ongoing video stream MAY need to yield to machinery control commands so they get received before their deadline 🥺 The price? The video stream quality gets temporarily degraded. Worth it, no cap 💖

The HPN working group is looking for mechanisms in the IPng protocols to provide for both of these classes in an integrated fashion ✨ An integrated services architecture cuts down on design and integration complexity by giving the mission critical system designer and application developer one uniform set of tools 🧚 And the integrated services architecture MUST be flexible and scalable, so new services can be added in the future with minimum impact on systems already using it 🌈 The HPN working group intentionally avoided naming particular mechanisms for solving these problems, so nobody gets locked into one solution 🫶

4.3 Mobility 🦋

okay so the HPN working group spotted two classes of mobility in the Navy mission critical environment 🚢 First, most platforms are themselves mobile. When they zip from port to port or flight deck to flight deck, it's important that they can still talk to a bunch of defense installations through a general infrastructure ✈️ Also, it's feasible that systems within a single platform MAY be mobile 📱 Maintenance and damage assessment requires large amounts of information at numerous locations on a platform. This information could possibly be made available through mobile terminals 💜

4.4 Multicast 🌸

Multicast transfer is a seriously critical IPng requirement for the Navy's mission critical systems 🌸 Picture a naval platform: tons of hosts (like workstations) connected through numerous subnetworks, all working different pieces of keeping the platform operational for its mission. So multicast is needed to share data that multiple hosts need at the same time. For example, on a ship, environmental data (roll, pitch, heading...) is needed by almost every system 🧭 Video conferencing MAY be used for chatting among operational personnel at multiple places aboard the ship, and it could also be used to talk with folks on other platforms or at shore facilities 📹 Both of these examples, plus a bunch of DoD and NATO studies, have highlighted the need for multicast functionality in mission critical systems 👑

One of the limiting factors with the present IP version 4 multicast is the OPTIONAL nature of this multicast, especially with respect to routers 😤 The use of tunnels, while enabling the initial deployment of multicast in the Internet, appears to limit its potential. The HPN working group believes the best approach to provision of multicast functionality is to consider it as a basic functionality to be provided by IPng 💜 In addition, sensible mechanisms are needed to control multicast traffic (i.e., scope control). Finally, support is REQUIRED to enable multicast functionality in IPng in areas such as group addressing and scalable multicast routing 🎀

4.5 Rapid Route Reconfiguration ✨

The HPN project is going to use very high bandwidth subnetwork technology 🚀 In the mission critical environment, one super important problem is putting a very low bound on how long it takes to identify a subnetwork problem and finish the necessary route reconfigurations ⏱️ The Navy's mission critical environment needs to be able to trade off bandwidth to get a short detection/reconfiguration time on subnetwork faults. A maximum bound on this time is felt to be less than 1 second ⚡

---

5. Additional considerations 💅

okay so this section is all the extra concerns from the mission critical environment that MAY impact IPng 🫠 The HPN working group felt these issues are important, but it wasn't clear how or whether IPng solutions need to accommodate them. It MAY suffice that IPng designers are simply aware of these issues, so they don't rule out reasonable solutions to these problems 🧚

5.1 Fault Tolerance 🫶

The mission critical environment is extra sensitive when it comes to fault tolerance 🥹 Any mechanisms that can fit inside the IPng protocol set, including routing and management, to support various levels of fault tolerance are desirable 💖 In particular, the following features SHOULD be supported: error detection, error reporting, traffic analysis, and status reporting 📊

5.2 Policy Based Routing 🧚

The HPN working group feels that there MAY be some uses for policy based routing within the Navy's mission critical systems 🔐 The primary interest is supporting a very capable security facility. Other uses discussed are keeping certain types of data on certain subnetworks (for multiply homed hosts), and providing automatic reconfiguration when particular subnetwork failures happen 🌸

5.3 Security 💜

Security is an important requirement for most Navy applications, so being able to design network functions to support security services is essential 🛡️ The following are several security services in particular that the HPN working group believes the network function SHOULD be able to support: rule based access control, labeling, authentication, audit, connection oriented and connectionless confidentiality, selective routing, traffic flow confidentiality, connection oriented and connectionless integrity, denial of service protection, continuity of operations, and precedence/preemption. On top of those, the network function SHOULD also support the security management of these security services. Key management in particular is of importance 🔑

Currently, the IPSEC of the IETF has several draft memos being considered to incorporate various security services in the network functions. It is of concern to the HPN working group that the IPng be able to support the concepts currently being developed by the IPSEC, and also provide the ability for the addition of future security services 👀

5.4 Time Synchronization 🌈

Time synchronization among the various components of mission critical systems is of vital importance to the Navy ⏰ It's desirable to be able to synchronize systems on multiple subnetworks via a network layer infrastructure. Some hooks for time synchronization can be envisioned in the network layer. However, the HPN working group feels that, as a minimum, efficient time synchronization algorithms MUST be able to function above an IPng infrastructure. For HPN systems, it's desirable that a time-of-day synchronization capability be supported of at least an accuracy of one microsecond among all hosts in a platform or campus network ⚡ The IPng protocols SHOULD NOT arbitrarily prevent this type of synchronization capability 💅

---

6. Conclusions 👑

A bunch of concerns specific to the mission critical systems targeted by the HPN working group have been identified 🎀 The HPN working group is interested in participating with the IETF in developing standards that would apply to mission critical systems 🤝 In particular, they're interested in developing multicast functionality, an integrated services architecture, and support for high performance subnetworks 🔥

7. References 📚

[1] HPN Planning Group, "Concepts and Guidance for High Performance Network (HPN)", Work in Progress, May 17, 1993.

8. Security Considerations 🚨

Security issues are discussed in Section 5.3. 💀 That's where the big stuff lives, so don't skip it.

9. Authors' Addresses 🎀

Dan Green
NSWC-DD
Code B35 NSWCDD
Dahlgren, VA 22448

Phone: (703) 663-1571
EMail: dtgreen@relay.nswc.navy.mil


Phil Irey
NSWC-DD
Code B35 NSWCDD
Dahlgren, VA 22448

Phone: (703) 663-1571
EMail: pirey@relay.nswc.navy.mil


Dave Marlow
NSWC-DD
Code B35 NSWCDD
Dahlgren, VA 22448

Phone: (703) 663-1571
EMail: dmarlow@relay.nswc.navy.mil


Karen O'Donoghue
NSWC-DD
Code B35 NSWCDD
Dahlgren, VA 22448

Phone: (703) 663-1571
EMail: kodonog@relay.nswc.navy.mil