Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [802SEC] some 802 history is captured here

When I started in 802 in 4Q1981, I was working on and looking for kindred spirits for doing hierarchical networks (10 and 100 Mb/s, support packet voice, consequently deterministic performance and the need to bridge between speeds).  The 20 Mb/s scope was definitely a problem for that vision.  The solution my colleagues and I at Burroughs were refining was a timed token ring.  

I'd originally invented the timed token protocol for an electrical CATV technology based bus.  So, being applicable to both ring and bus, I proposed the protocol to the Token DLMAC (soon to become 802.4 and 802.5).  It was accepted for a few weeks.  Then I got a rude awakening to the politics of standardization at the hands of IBM and others, with Token Ring changing back to being based on the IBM Zurich protocol, but 802.4 retaining the timed token protocol.

Sperry showed up in 802.5, but found it inhospitable for various reasons, one being the speed issue.  They were though very open to any MAC protocol that would work for their requirements (determinism, speed and reach).  

So, Sperry seriously engaged in X3T9.5 work.  When I arrived in X3T9.5 in 3Q1983?, I think they had already tagged their effort as FDDI.  The name fit with the ongoing work on LDDI which as I recall was based on DEC's CI VAX interconnect.  Sperry quickly bought into using the timed token protocol for FDDI, and so a Burroughs MAC was hooked up with a Sperry PHY and the project got traction.  DEC was still around because of LDDI, and they, HP and others quickly got sucked into the work.  Chip companies joined and as well as too many other companies, and progress ground to a snails pace complexity increasing (especially in FDDI SMT).  Sad considering the capability of many of the individual contributors involved.

Just as a tangent, some others at Burroughs had added an isochronous interrupt capability to the timed token protocol and I proposed the whole package at the beginning (which probably would have met the later FDDI-II requirements, but others didn't see any isochronous need at that time).  When re-proposed a few years later for FDDI-II, the isochronous folk didn't like it because it "complicated" slot interchange at the bridge (burst rate of the isochronous data was much faster than the approach for FDDI-II).  IBM TJ Watson independently invented a very similar interrupt and proposed it later for FDDI follow-on, and some might remember another independent invention and abortive (post gigabit?) CFI for adding similar interrupt capability to Ethernet.

The isochronous capabilities that spawned FDDI-II were the same thing driving 802.9.  Both had TDM bandwidth partitions for packet and isochronous data, but the latter using Ethernet as the packet MAC protocol rather than the timed token protocol for FDDI-II packet data.

802.6 also played with isochronous data ideas, so I don't recall our history for focus on variable size packets being as pristine as Geoff recounted.

(Committee shopping to find a place to do your standard isn't new.)  

The scope of X3T9.5 from my perspective was clearly backend networks (now would probably be called a Data Center network).  There was a need for long links in back-end to remote disks and tapes outside the computer room (archival, secure survivable backup, etc.)  So significant link lengths were defendable for back-end applications, even though it allowed most everyone to see FDDI's potential as a backbone LAN to connect 802 LANs.  (Where it found its time in the market before Fast Ethernet squashed it.)

As Ethernet folk know, X3T9.5 also developed TP-PMD for FDDI, and they chose to take the years of work on FDDI PHY, FDDI PMD, and FDDI TP-PMD as the basis for the physical layer for Fast Ethernet (100BASE-FX and 100BASE-TX).  IEEE 802.3 consequently got faster time to market as well as establishing the pattern for multiple future Ethernet speed increases.

So, the 802/X3T9.5 inter-relationship was pretty active.  And more than a few people participated in both committees.

Pat, I don't remember the metro reach of 802.6 being a turf controversy with X3T9.5, but I do remember it being a scope controversy.  The original 802 charter included reach assumptions, which 802.6 clearly intended to exceed.  I think all the controversy was internal to 802.  By then it was pretty clear X3T9.5 was really developing a LAN (even though it still had back-end network requirements), and consequently there was no interest in starting a turf fight with 802 from the X3T9.5 side.  

I've occasionally joked that I might be the most promiscuous 802 participant (802.4, 802.5, 802.6, 802.9, 802.11 and finally with gigabit settling in 802.3).  While prior to that on the X3 side also participating in FDDI, FDDI-II, FDDI Follow On, and HiPPI.  With all those committees, details can get blurred, so if others remember it differently we might be able to clarify things together.


-----Original Message-----
From: ***** IEEE 802 Executive Committee List ***** [] On Behalf Of Pat Thaler
Sent: Monday, July 27, 2009 1:31 PM
Subject: Re: [802SEC] some 802 history is captured here

A bit of interesting history regarding the evolution of 802 scope is hinted at by the mention of X3 at the end of the letter in the pdf.  X3 was working on FDDI in X3T9.5 and the upper limit of 20 Mbps for 802 networks was to keep us as LANs and leave higher speed MANs to them. I recall an exciting EC meeting (back when I was in the peanut gallery) because the PAR for 802.6 was being considered which stepped on their toes. That added MAN to our scope and eventually 802.3 also discarded the 20 Mbps upper limit for its work. 

-----Original Message-----
From: ***** IEEE 802 Executive Committee List ***** [] On Behalf Of Paul Nikolich
Sent: Monday, July 20, 2009 11:06 AM
Subject: [802SEC] some 802 history is captured here

This email is sent from the 802 Executive Committee email reflector.  This list is maintained by Listserv.

This email is sent from the 802 Executive Committee email reflector.  This list is maintained by Listserv.

This email is sent from the 802 Executive Committee email reflector.  This list is maintained by Listserv.