Terabit Routers

Rocket Boy writes "I was perusing the news and came across this sucker. The specs on the thing look outrageous. Heck, the whole thing looks outrageous. 2.5-5.6TB/S speed, Supports 2240 OC-48 or 560 OC-192 connections. " You can download a lot of po.. I mean play a lotta qua... I mean read a lot of slas... I mean.. work. You could do a lot of work with that.

Comments

Not news

[apilosov]

Pluris was doing it (or claimed to) since '97. They used hypercube for internal switching fabric configuration, and I believe used wireless very-short-distance (1m) links to reduce amount of wiring. Vadim Antonov was the brains behind it, he left the company in '98. Pluris didn't release anything, and their website (www.pluris.com) seems down. Vadim also is person who named major russian UUCP host as 'kremvax', developed/ported unix for soviet PDP-11 clone (called DEMOS), founded first russian ISP (also called DEMOS), architected much of SprintLink, etc.

Avici on other hand doesn't really say much about how their fabric is done, but I imagine it won't be crossconnect (2000x2000 wires...ugh). Probably some sort of hypercube or selective mesh. I wish Avici was as forthcoming with technical details as Pluris. :)

I'll believe it when they build one

[Salamander]

Saying something is "scalable to X terabits" doesn't mean much unless you specify what an "X-terabit router" really has to do. A big $#@! pile of PCs is "scalable to X terabits" if they're all handling totally independent traffic flows and never have to talk to one another, for example. Probably the most useful measure is "cross-sectional bandwidth" (i.e. the amount of bandwidth you'd have to remove to partition the nodes into two or more isolated subsets) as used for measuring intramachine interconnects, but - alas - Avici doesn't give us that information.

There's also this little issue of balance. Nothing scales perfectly, and often you don't know where the bottleneck will be until you build one. Sure, if you add up all the links maybe you get up to X terabits, but maybe node-internal contention for some resource limits you to X/100. Of course, this never stopped marketing types from acting as though their machine/link/router would be the first in the history of computing to scale perfectly.