Cisco Reveals Its $500 Million Router

Whitecloud writes "After 4 years of development and $500 million in costs, Cisco have a new router: the CRS-1, or Carrier Routing System. Cool features include a 40 gigabit-per-second optical interface, and the ability to cluster the boxes to act as a single router. retail starts at $450,000. Video available here." Update: 05/26 13:55 GMT by T : Sorry; I missed the previous mention of this device.

Re:interesting math

[Timesprout]

Can we really use 10,000 of these things

Maybe not right this second but demand for bandwith is only going to grow, and probably more rapidly than currently, for the foreseeable future as the entire world becomes digitized and goes online

Re:interesting math

[DaHat]

Quite true, but I would expect that the IP created for the HFR (Huge Fast Router) could be applied to other Cisco products in the near future with a higher product margin.

Re:interesting math

[liam193]

Can we really use 10,000 of these things? That's a whole heck of a lot of throughput...

Yeah, I see where your going. It's just like memory in PCs, 640 of these is all anyone could ever use.

Re:interesting math

[grub]

If they spent 500M on this, and they sell for 450T, and they have a 10% profit margin (unlikely, but it's a round number) then they'd need to sell +10,000 of these boxes to make a profit.

Right, but the target market for these boxes will likely have "Cisco" logos all over their networking racks. Even if they don't make money on this line, they won't have a competitor (Juniper, Nortel, etc) getting a foothold in the data center.

10 years?

[bo0ork]

Watching the video, they proudly proclaim that this product will allow a service provider to do their thing for the next ten years. Yeah. Right. With the way bandwidth-for-the-consumer is going, the ISP's are going to need petabits of routing capacity in ten years, not gigabits.

Re:interesting math

[naelurec]

It probably won't be profitable if you look at it this way. However, a lot of the R&D to develop this router will find its way into a wide range of other products.

Re:/. should get one.

[Sheepdot]

Naw, I'm just noticing that there are a lot more posts about products and services on /. rather than the just the technology surrounding them as of late.

Re:interesting math

[afidel]

Cisco's margins are more along the lines of 50-60% depending on the product line. I know because the wireless division was being dumped on for only having ~40% margins. Then the bubble burst and other divisions suddenly had almost no sales so those margins on increasing sales started to look not so bad =)

Re:interesting math?

[OzeBuddha]

Well actually, assuming every one of these routers made Cisco $450K in pure profit, and given that they have spent $500M in development, they would only need to sell just over 1,100 of these things to cover costs.

Note that the post states that the routers start at $450K and also note that the router itself must cost something to make apart from the R&D costs, so the number of routers that Cisco must sell in order to make a profit is probably somewhere closer to 2,000 or 3,000. Perhaps they do not plan to make a profit initially, believing that the technology that they have now developed will lead to more optical switching products that will make them mega bucks in the future..

Don't forget that the entire worldwide demand for computers was only ever supposed to be a handful..

I'm sure that we will find something to do with multiple 40Gbps routers..

Multi-player Network video Dance Dance Revolution [ncsu.edu] EXTREME [gamer-talk.net] deathmatch anyone?

Re:Switches

[arivanov]

No. It is jumping in a mature market, conceding to the market demands and playing by the market rules. In fact it is the first IP vendor to do so.

The biggest demand and the main objection to IP by all big telcos since the first days of the Internet has been that you cannot interface routers directly into the provisioning backend and that you have to keep highly qualified expensive staff to run it instead of paying a fraction of that for backend software and coasting on it for 7-9 years.

Cisco is the first one to comply with this demand from the IP vendors, but not the last one. In fact Juniper is about to follow, others will also jump on the bandwagon.

It is the first router to have an XML/SOAP interface that can be plugged into the provisioning/maintenance system via an industry standard for interfacing large systems so you no longer need to employ a bunch of CCXX-es to bang on keyboards. In fact it is what carriers have been asking to use MPLS for a while now and similar to what the ITU would have forced down everyone's throat anyway.

This also means that any CCXX that is not accompanied by computing background has just dropped in value and will continue to drop in value as Cisco releases the new IOS to other devices accompanied by tools.

I can understand them doing it. Their revenue from certs has nearly leveled now after that mad rush at the end of the boom. It is time to pick up a new revenue stream in the form of upgrades to Cisco Wors (favourite oximoron) and interfacing to carrier systems.

Re:Switches

[sphealey]

The biggest demand and the main objection to IP by all big telcos since the first days of the Internet has been that you cannot interface routers directly into the provisioning backend and that you have to keep highly qualified expensive staff to run it instead of paying a fraction of that for backend software and coasting on it for 7-9 years.

First you are going to have to convince me that the telecomm carriers have a "provisioning back end" that consists of anything more than a bunch of grade school kids in Bangledesh looking things up in ledger books. No - scratch that - back before 1980 when my county used ledger books to keep tax records for 1,000,000 properties, that system was a lot more efficient than telecomm provisioning is today.

45 day notice for a new circuit, and they still get it wrong 80% of the time or more, requiring two or three 12 hour days on the phone and a complaint faxed directly to a Vice-President's office before anything gets done. Is that a system that should be catered to?

sPh

Re:Switches

[sphealey]

Your milage cetainly will vary!

Things seemed to be getting better for a while. Back in the 1998-1999-2000 time frame I had new circuits in in 3 weeks (with only 1 day on the phone!), and expansion of existing networks sometimes as fast as 10 days.

But lately it has been 30-45 days, with the occasional 90 day @#@#!$-up. And no one at the telecomm companies seems to know what is going on.

sPh

Re:This would be interesting..

[Anonymous Coward]

I can see the DOD and Tier-1 ISPs having an immediate use for this. Especially with the move to IPv6, you are going to see a lot more 'centralized' routing because of the hierarchical structure of networking with IPv6. They are aiming to reduce the size of the default router table, and that means that the core routers will have to each handle a lot more traffic than before, that's not even counting future growth. For IPv6 to succeed, the top-end routers are going to have to become much more powerful and efficient.