RFC 1222

Advancing the NSFNET Routing Architecture 🦋

okay bestie, vibe check first ✨ this is RFC 1222, from the Network Working Group, dated May 1991. H-W. Braun (San Diego Supercomputer Center) and Y. Rekhter (IBM T.J. Watson Research Center) wrote it. It's informational, which means it's giving "here's some info for the Internet community," not a standard 💅 The title block doesn't list anything it obsoletes or updates, so no drama there 🙈

---

Status of This Memo 📜

so this RFC is all about improving the NSFNET routing architecture so it can offer a more flexible interface to the Backbone clients 🌸 it provides information for the Internet community. it does not specify an Internet standard. distribution of this memo is unlimited 🫶

Introduction 🧚

this memo tells the history of NSFNET Backbone routing and lays out two suggested phases for how the Backbone's routing interface could keep evolving 🌈 the goal is a more flexible interface for NSFNET Backbone service subscribers, through an attachment option that's simpler and lower-cost than the current one 💖

Acknowledgements 🥹

the authors want to thank Scott Brim (Cornell University), Bilal Chinoy (Merit), Elise Gerich (Merit), Paul Love (SDSC), Steve Wolff (NSF), Bob Braden (ISI), and Joyce K. Reynolds (ISI) for reviewing the memo and giving constructive comments 🫶✨

---

1. NSFNET Phase 1 Routing Architecture ✨

phase 1 was the first NSFNET Backbone, running on a 56Kbps infrastructure with routers based on Fuzzball software [2] 🍓 the Phase 1 Backbone used the Hello Protocol for interior routing. at the periphery, the client networks were usually connected through a gatedaemon ("gated") interface, which translated between the Backbone's Hello Protocol and the interior gateway protocol (IGP) of the mid-level network 🔥

mid-level networks mostly used the Routing Information Protocol (RIP) [3] for their IGP 🦋 the gatedaemon system acted as the go-between for the Hello and RIP worlds 🧚 the overall look was that the Backbone, mid-level networks, and campus networks formed one single routing system where info was freely exchanged 👑

network metrics got translated across all three levels: the backbone, the mid-level networks, and the campuses 🎀

once the gatedaemon arrived, sites could add filtering based on IP network numbers 💅 and each individual site's staff controlled that filtering 🌸

once specific network routes were learned, the infrastructure forwarded metric changes throughout the interconnected network 😭 the end result: a metric fluctuation on one end could permeate all the way to the other end, crossing multiple network administrations 💀 the frequency of metric fluctuations inside the Backbone itself went up even more when event-driven updates (e.g., metric changes) were introduced ✨ damping later lessened how often those updates happened, but the overall routing environment still seemed pretty unstable 🥴

given that only limited tools and protocols were available for engineering the flow of dynamic routing information, it was pretty easy for routing loops to form 🙈 this got amplified as the topology became more fully connected without insulating routing components from each other 🌀

all six nodes of the Phase 1 Backbone were located at client sites, specifically NSF funded supercomputer centers 🧠

2. NSFNET Phase 2 Routing Architecture 💅

phase 2 ran on T1 (1.5Mbps) lines and focused on the lessons learned in phase 1 📚 the big result was a strong decoupling of the IGP environments of the backbone network and its attached clients [5] 🫶 each administrative entity could use its own IGP in any way that made sense for its network 👑 the interface between the backbone and its attached client was built with exterior routing, initially via the Exterior Gateway Protocol (EGP) [1,4] ✨

EGP improved routing isolation in two ways 🦋 first, EGP only signals up/down transitions for individual network numbers, not the fluctuations of metrics (with one exception: metric acceptance of local relevance to a single Nodal Switching System (NSS) only, for inbound routing information, when there are multiple EGP peers at an NSS) 🤓 second, it allowed engineering of the dynamic distribution of routing information 🌈 that means primary, secondary, etc. paths can be determined, as long as dynamic externally learned routing information is available 🗺️ this lets you create a spanning tree routing topology that satisfies the constraints of EGP 🌳

the pre-engineering of routes happens through a routing configuration database that's centrally controlled and created, and then individual configuration info gets distributed to all the NSFNET Backbone nodes 🧚 a computer-controlled central system makes sure the database is correct before it's distributed to the nodes 🔐

all nodes of the 1.5Mbps NSFNET Backbone (currently fourteen) are located at client sites, like NSF funded supercomputer centers and mid-level network attachment points 🍓

🎀 part 1 ends here, part 2 is coming with Section 3 and the near-term routing alternative 🎀

5. Long-Term Alternatives ✨

okay so bestie, let's talk about the future 🔮 as the NSFNET Backbone technology keeps evolving, you can picture the dividing line between the Backbone and the service subscribers sliding toward the "C-NSS cloud" 🌥️ which means the NSFNET IGP gets confined to the C-NSS, while the E-NSS becomes a full participant in the service subscriber's IGP. it's giving clean boundaries 💅

the big prerequisite here is being able to actually manage the physical medium connecting a C-NSS and an E-NSS when two different administrative entities are involved (that's the NSFNET Backbone provider AND the service subscriber, fr fr 🫶). it also has to be manageable enough to be as easy to use as an Ethernet interface, working as a well-defined demarcation point. no cap, that's a high bar 🎀

the evolution of the Point-to-Point Protocol 🦋 and a way way better capability for managing serial lines through standard network management protocols will definitely help 💖 but this may not be the whole answer 😭 because a bunch of different equipment sits on serial lines, which makes it hard to isolate a hardware problem. similar issues might pop up for future demarcation interfaces to Internet infrastructure too (like SMDS interfaces 🌸).

in summary, there's an opportunity to simplify management, administration, and the exchange of routing information by collapsing the number of physical entities involved. main character energy ⭐ literally the goal 👑

---

6. References 📚

[1] Mills, D., "Exterior Gateway Protocol Formal Specification", RFC 904, BBN, April 1984.

[2] Mills, D., and H-W. Braun, "The NSFNET Backbone Network", SIGCOMM 1987, August 1987.

[3] Hedrick, C., "Routing Information Protocol", RFC 1058, Rutgers University, June 1988.

[4] Rekhter, Y., "EGP and Policy Based Routing in the New NSFNET Backbone", RFC 1092, IBM T.J. Watson Research Center, February 1989.

[5] Braun, H-W., "The NSFNET Routing Architecture", RFC 1093, Merit/NSFNET, February 1989.

[6] Braun, H-W., "Models of Policy Based Routing", RFC 1104, Merit/NSFNET, June 1989.

[7] Lougheed, K., and Y. Rekhter, "A Border Gateway Protocol (BGP)", RFC 1163, cisco Systems, IBM T.J. Watson Research Center, June 1990.

[8] Almquist, P., "Requirements for Internet IP Routers", to be published as a RFC.

---

7. Security Considerations 🔒

okay bestie, this one's short but it's SO important to be honest about 😭 this memo does NOT discuss security issues at all 💀 so if you're looking for a security section, it's not here, periodt 🙈

---

8. Authors' Addresses 🎀

Hans-Werner Braun San Diego Supercomputer Center P.O. Box 85608 La Jolla, CA 92186-9784

Phone: (619) 534-5035 Fax: (619) 534-5113

EMail: HWB@SDSC.EDU

Yakov Rekhter T.J. Watson Research Center IBM Corporation P.O. Box 218 Yorktown Heights, NY 10598

Phone: (914) 945-3896

EMail: Yakov@Watson.IBM.COM

the end, bestie 🌈💜 that's the whole NSFNET routing story, and these two legends wrote it 🧚