Cisco Barges Into the Server Market

mikesd81 was one of several readers to write in about Cisco's announcement of what has been called Project California — a system comprising servers made from 64-bit Intel Nehalem EP Xeon processors, storage, and networking in a single rack, glued together with software from VMWare and BMC. Coverage of this announcement is everywhere. Business Week said: "The new device, dubbed Project California, takes servers into new territory by cramming computer power into the very box that contains storage capacity and the networking tools that are Cisco's specialty. Cisco's approach could help companies use fewer machines — saving money not only on hardware, but also on power and IT staffing — in building data centers. ... Cisco is well-girded to take this step. It has more than $30 billion in cash, more than any other tech company. The company is moving into no fewer than 28 different markets, including digital music in the home and public surveillance systems." The Register provides more analysis: "Microsoft is, of course, a partner on the California system, since you can't ignore Windows in the data center, and presumably, Hyper-V will be supported alongside ESX Server on the hypervisors. (No one at the Cisco launch answered that and many other questions seeking details). ... The one thing that Cisco is clear on is who is signing off on these deals: the CIO. Cisco and its partners are going right to the top to push the California systems, right over the heads of server, storage, and network managers who want to protect their own fiefdoms."

Comments

Why use bleeding edge intel chips?

[ShooterNeo]

I have to ask : why Nehalem EP Xeons? Those are the absolute bleeding edge chips that Intel manufactures, and as such as the most expensive by a significant margin. Newegg doesn't even have the chip listed on their website, yet carries 91 different server CPU models. While space inside the data center does cost money, and so does electricity, is it really so expensive as to be worth paying for a chip that is probably 10 times as expensive per MIP as cheaper alternatives? The motherboards are more expensive as well, especially when you factor in the huge markup for server grade parts.

The only advantage of the Nehalem is that it is SLIGHTLY faster per processing thread, but networking is usually an "embarassingly parallel" problem.

Re:Why use bleeding edge intel chips?

[Zeio]

One of the main things that would make Nehalem attractive is a few things for blade servers. The process 45nm moving on to 32nm later will provide the smallest footprint while not giving up any CPU power. Also, the Nehalems (Bloomfield and Core i7) that I have tested don't seem to offer much in terms of better performance, but the power usage is considerably lower. Also, FB-DIMMS (and DDR2) were a bit too consumptive of power, the newer memory technology is an attempt to reduce power consumption. Also, the CPI (formerly CSI) offers the Intel CPUs a Hypertransport-like interconnect system allowing system builders to scale. The footprint of 16+ core systems with Nehalem will be far smaller than the previous generations of Xeon MP processors.

I think the idea is more memory, more CPU processing power, less power and heat and scalability with a given architecture.

I noticed in my testing the L1 Data and instruction caches (32KB per core) in the Core i7 is one cycle more latent than the core2 (4 cycles vs 3), the L2 cache (which is 256KB per core rather than the 4-6MB per two cores in Core2) is faster, down from 17 cycles to 12 cycles. With this boost in L2-speed came a cut of 3.75MB-5.75MB in size. The way they mitigated that loss was to give a "large" 8MB L3 cache that runs on a slower clock. This new system, along with a hyper-threading implementation that, unlike the previous one, seems to genuinely enhance performance in nearly every test, allows Intel to make top-performing chips (see CPU 2006 @ spec.org for the latest results) that scale better via QPI, use less power and fit into smaller spaces than previous chips.

See:
Sorted SPEC CPU2006 Integer and Floating Point

CINT2006
Hardware Vendor System Result Baseline # Cores # Chips # Cores Per Chip Published Disclosure
1) YOYOtech Fi7EPOWER MLK1610 (Intel Core i7-965) 36.0 32.5 4 1 4 Jan-2009
2) ASUSTeK Computer Inc. ASUS P6T WS PRO workstation motherboard (Intel Core i7-965 Extreme Edition) 35.2 31.5 4 1 4
3) ASUSTeK Computer Inc. Asus P6T Deluxe (Intel Core i7-965 Extreme Edition) 33.6 30.2 4 1 4 Nov-2008
4) ASUSTeK Computer Inc. Asus P6T Deluxe (Intel Core i7-940) 30.8 27.8 4 1 4
5) Dell Inc. Dell Precision T7400 (Intel Xeon X5492, 3.40 GHz) 30.2 27.6 8 2 4

CFP2006
Hardware Vendor System Result Baseline # Cores # Chips # Cores Per Chip Published Disclosure
1) ASUSTeK Computer Inc. ASUS P6T WS PRO workstation motherboard (Intel Core i7-965 Extreme Edition) 39.3 37.4 4 1 4 Feb-2009
2) YOYOtech Fi7EPOWER MLK1610 (Intel Core i7-965) 35.7 33.6 4 1 4 Jan-2009
3)ASUSTeK Computer Inc. Asus P6T Deluxe (Intel Core i7-965 Extreme Edition) 33.6

If you can't make the sale, move up the org chart!

[dweller_below]

Yep. That's the Cisco I know and loath. If you can't convince the literate, just move up the org chart.

Years ago, at my institution (150+ buildings, about 15K active IP addresses,) we did a cost analysis of our Cisco addition and decided that it was unnecessary. We could do everything we needed with cheaper, commodity devices.

So, for the next couple years, all upgrades/replacements were to simpler structures. To non-proprietary protocols. And to non-Cisco equipment. We have been Cisco-Free for about 4 years.

The hardest part was beating off the attacks from Cisco Sales. These attacks were vicious. They lied (even more than usual for Cisco sales droids.) They tried their best to discredit us. First they approached the head of IT. Then the VP for Business. Then the president.

Finally, they went to the Board of Regents. They said we were incompetent. They said our actions were endangering the future of our institution. Fortunately, the Regents decided to let us try it.

It has worked out great for us. Our capability is up. Our reliability is way up. Our security is up. Our costs are down (about 1/2 the price of equivalent Cisco.)

But, it only happened because upper management was willing to trust us. I get the impression that most management would fold under the pressure we saw.

Miles

Re:They go for the "soft" target

[OddlyMoving]

This is very true. I am currently evaluating a forklift upgrade of one of my POPs, and we're looking at the Cisco vs. Juniper proposition.

While I'm a VP level operational head at an ISP, the Cisco rep told me straight out that he doesn't typically engage technical people like me when he comes in. He typically talks to the C level people, and it shows, because he's not keeping up with the Juniper rep. The Juniper team has already put me in front of many technical product development people, and the depth of the conversations have been truly refreshing. I'm feeling more and more comfortable with going Juniper as the days go by.

All smoke and no fire....

[trims]

OK. My bias up front - I work for Sun.

That said, there were several pre-Cisco-announcements from HP, IBM, and Sun about how the California system is a no-go. Admittedly, they're the competitors for Cisco, but after having looked at the existing rack blade/switch systems from those three vendors, I really don't see any difference worth mentioning from current product lines.

Here's some thoughts:

• IBM and Sun make much more Open systems, able to run a wide variety of vmWare, Linux, Solaris, and even AIX on all sorts of hardware (SPARC, POWER, PPC, all sorts of AMD and Intel x64). Their systems are much more flexible and honestly, much more powerful overall in what can be accomplished.
• HP has much of the HW flexibility of Sun and IBM, plus the leading management tools.
• Cisco has no clue as to how to run a systems support organization, which, frankly, is considerably different than running a network hardware support organization. The other big three have decades each in doing this kind of thing.
• Sun in particular has extremely competitive pricing. HP and IBM are slightly more expensive, but nothing compared to the margins Cisco charges. So, exactly WHAT are people going to get for the 20-40% premium Cisco is charging over IBM?
• Even for the Virtualization craze, building a completely proprietary solution flies in the face of what everyone else in the industry is doing: commoditization.
• Cisco doesn't have integrated solutions. All the others provide storage, network, and compute integration with large, well-trained Professional Services orgs. Cisco has CCIEs in piles, but what do they know about anything but network gear?

Overall, this looks like a stupid move. I realize that Cisco needs to look for more revenue streams in the face of the commoditizing of most network gear, but this seems like an '80s solution to a 2010 problem.

-Erik