Cisco's big data center plans: Assessing winners and losers rack by rack

Cisco's big data center plans: Assessing winners and losers rack by rack

Summary: Cisco on Monday rolled out its long-awaited vision of data center architecture, including multiple partners, a wire-once approach to link computing, virtualization, storage and networking and a call to lower costs. Let's assess the winners and losers of this rack-by-rack data center battle.


Cisco on Monday rolled out its long-awaited vision of data center architecture, including multiple partners, a wire-once approach to link computing, virtualization, storage and networking and a call to lower costs. Let's assess the winners and losers of this rack-by-rack data center battle.

As background, Cisco's Unified Computing System aims to unify components of the data center into one footprint. The goal is to cut the total cost of ownership by 20 percent in capital expenses and up to 30 percent in operating expenses.  

Also see: Tom Steinert-Threlkeld's coverage of the press conference, Cisco statement, Cisco on Twitter here and here, Techmeme and more resources.

How did Cisco reach those savings figures? A Cisco spokesman said capital expenses will fall because fewer switches, adapters and cables are needed. Cables represent 15 percent of data center capital expenses. Cisco's architecture only has one cable going in. The set-up is also expected to lower power, cooling and labor costs. Meanwhile, one management application can manage the entire system compared to seven with legacy systems. 

As expected, Cisco also launched blade servers that are based on Intel's Nehalem processors. Cisco's design aims to create a single system that wires the blades and consolidates networking functions (local area, storage area and computing networks). 

Here's the setup via Cisco (at a glance and overview whitepaper):

Cisco's has taken a holistic approach to data center architecture and the ramifications could be big over time. However, the effects won't be felt overnight. Cisco noted that it doesn't expect a rip and replace of existing data center infrastructure in favor of its Unified Computing System. That said, Cisco will plan to win rack-by-rack as data center assets are replaced.

Cisco CEO John Chambers told customers that he realizes that they will buy "steps at a time." He also argued that Cisco's architecture is very appealing to greenfield implementations. "The data center is evolving. It had no choice but to change. Silos don't work anymore," said Chambers. 

Here's a look at the potential winners and losers:

Winner: Data center managers looking to find more productive uses for cable jockeys. For all of the moving parts of Cisco's Unified Computing Center announcement for IT execs it really boils down to two words: Wires and cables. All of those wire and cables cost you money. Cables equal labor costs and Cisco's new architecture has one wire going in. 

Instead of this:

You have this:

Cisco is trying to bring sexy design to the data center. Jim Grant, BMC's senior vice president of strategy and corporate development said the better wiring means IT managers can "wire once and reconfigure many in logical way rather than physical."

Loser: Competitors that will need to bolster their data center architecture plans. In a blog post, Cisco said that its data center move is to add value and not take on entrenched players.  

Cisco said:

This is not the ‘Clash of the Titans’ or us ‘coming after HP or IBM or Dell’. Some companies may choose to join us in this market, others may continue to operate the status quo.  We are certain over time we will not be alone in this market.

But Cisco's design work here is substantial. As Cisco gains its footing and begins rack by rack warfare customers are going to start asking about the Unified Computing System approach. Other data center rivals will have to start answering questions about integration between functions, architecture and the potential savings. If anything just the buzz associated with Cisco's move can dictate the conversation. 

Meanwhile, Cisco's partner lineup---SAP, NetApp, EMC, Wipro, Microsoft, Novell, Oracle and a bevy of others---means that customers will ask what companies are not on board, said Chambers.

Also see: Cisco seeks a new market that stops the madness

Cisco: Beyond the data center

Gallery: Cisco moves into blade servers

Winner: BMC Software. BMC and Cisco collaborated on the management software portion of the Unified Computing System. This deep integration means one app can manage the system. For BMC, the Cisco partnership "is preferential," said Grant. BMC, which is already entrenched in data centers, will find itself in all of those next-gen data center conversations. 

Winner: Intel. The chip maker gets its Nehalem processor to ride shotgun on the Cisco news. The positioning of Nehalem and next-generation data centers will be hard for AMD to match. 

Loser: Hewlett-Packard. There are a few items to ponder here. On the hardware side of the equation, HP may see blade server margins squeezed. HP is also battling back with its networking gear. But there's another hook here that may be overlooked: HP's own automation and management software. HP acquired Opsware and the capabilities are similar to what BMC offers. Now that Cisco and BMC are partners on sales calls it could be more difficult for HP's Opsware. 

Winner: VMware. As part of Monday's announcement, Cisco and VMware announced an original equipment manufacturer agreement. VMware, which is under fire from the likes of Microsoft, Red Hat and Citrix, gets another channel to push its virtualization platform. Cisco is a nice partner to have on board. “This is perhaps the only evolutionary road,” that is feasible right now, said Paul Maritz, the chief executive of VMWare, that could lead to “revolutionary” new forms of computing.

Loser: Brocade. Brocade made a big move to acquire Foundry Networks to push its own data center vision. Cisco's announcement will put pressure on Brocade to position its version as an alternative. Meanwhile, EMC is another big Cisco partner on its Unified Computing System as is NetApp. 

Brocade responded:

"Cisco's approach to unified computing is not revolutionary. Many companies with extensive experience, including Brocade, in solving complex data center issues are already working on solutions. Cisco's approach is likely to be very capital intensive up front, which will be a major obstacle in light of today's global economy."

Brocade also noted:

"A dynamic and virtualized data center holds the promise of many compelling benefits for end-users including increased server utilization, decrease in power footprint and more efficient operations in general. However, achieving this goal is a complex challenge that can be best tackled by a broad ecosystem of industry partners and not based on a proprietary, singular architecture of one company."

Winner: Cisco. The company has the message and the business case to make a go of its Unified Computing System. In many respects, the company has changed the data center conversation and managed to crowbar itself into more discussions as infrastructure is replaced.

Topics: Storage, Cisco, Data Centers, Hardware

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • I would argue...

    that Cisco could be a loser by locking customer's into the BMC tools instead of using a open architecture for management.

    [i]Insanity: doing the same thing over and over again and expecting different results. Albert Einstein [/i]
  • Cisco always overprices....

    everything. Why would this be any different? I doubt that anyone will lose here except those that decide to go with Cisco.
  • Not good for Cisco

    For Cisco to make a server.. great. You are cutting into a 3 or 4 way market (Dell, IBM, HP).

    For HP to get ticked of and start cranking out routers? I see that as a much bigger problem. There are a lot of Asian contract manufacturers with the expertise to build them. Slap on a polished Open Source OS (which HP has programmers to do).. and that spells trouble.

    HP wouldn't even have to take the thing seriously - just enough to make a mid-range model to send Cisco a message.
  • RE: Cisco's big data center plans: Assessing winners and losers rack by rack

    If you have ever done any kind of work with Cisco you would know they are all about locking you into their proprietary, espensive solutions. You can't user the terms "Cisco" and "standards" in the same sentence with a straigh face as they are the only vendor in a silo in many cases. Their NAC solution comes to mind as a classic example of proprietary, lack of standards support, and WAY overpriced!

  • Cisco marketing muscle getting the higher prices?

    Can Cisco's nameplate and marketing arm drive the customers to pay a premium for an integrated solution? I have no doubt that Cisco blades will cost more than similar HP/IBM units. They will try to hide the cost in a bundle of stuff like VMware and mgt tool licenses.

    HP is going to lose all the way around, because they will either lose the sale or lose the profit because they discounted below Cisco.
  • RE: Cisco's big data center plans: Assessing winners and losers rack by rack

    Cisco does not have any advantages compelling enough to take market share. So Cisco will probably remain a niche player & eventually decide to bow out of the server market. IBM is probably the strongest because it has whole variety of CPU's.
  • RE: Cisco's big data center plans: Assessing winners and losers rack by rack

    IMO, HP is the loser here. When they stepped into the networking market, they brought no innovation. Now Cisco enters the compute market with a real innovation. I agree it's not a fit for all environments, but I can see where it makes sense for greenfield operations. The Juniper/IBM relationship will need to really step up the R&D plate. Based on history, Juniper will follow Cisco's lead in 24-36 months.
    • No innovation? Are you kidding?

      "[i]IMO, HP is the loser here. When they stepped into the networking market, they brought no innovation. [/i]"
      HP has been in networking from their Roseville California campus since the 1980ies, and co-developed the twisted pair Ethernet cabling technology most people use every day (and Ethernet 802.3 came out of DEC labs with some assistance from Xerox and Intel). They've offered a life-time warranty on their networking gear (EtherTwist, then AdvanceStack, then Procurve) since the early 90ies and their stuff is a 1/4 the price of Cisco, [i]and[/i] you don't have to pay yearly support for updates. If you open an HP switch and compare it to the equivalent cisco, the HP is all custom ASICs and the Cisco looks like someone built it from spare components. HP offered multi-link trunking (early-mid 90ies; I remember working on a pilot project) with load balancing long before Cisco would even consider the topic, and while it lost in the long run 100VG was light years ahead of 100Base-T (built in QoS, less expensive networking gear, 25% of radiated interference of 100Base-T on [i]cat 3[/i]; 100Base-T requires Cat 5 at a minimum. Unfortunately it was a case of Beta vs VHS). The Fibre Channel switches that Brocade sells today evolved from a design that came out of HP Labs in the early 90ies. IMHO, HP should have bought Cisco 20 years ago when they were considering it.
  • RE: Cisco's big data center plans: Assessing winners and losers rack by rack

    I like the last slide, the one entitled "Unified Computing System launch". Is it just me, or do they win the b@llshit business lingo award? I'm sorry, I can glean no relevant information from that piece of marketing, short of whoever wrote it went to business school.

    Cisco is smoking rope if they think they are going to put out superior blades and software against HP, Dell, and IBM. Servers, storage, and management software has been those companies core services for years, and Cisco thinks they will show up to the party and make great sales in an economic downturn?

    And having used Cisco hardware and software in my previous datacenter, I can say we spent a lot of money and I was not impressed. They really should focus on improving their core products, user experience, and software line before they try to dominate another market.

    They're going to burn a lot of cash and likely end up with a mediocre offering. Another great visionary idea from an clueless executive, no doubt.
  • I suspect they will get some traction...

    ...but this type of "wire-once" solution is nothing new, both HP and IBM have offered solutions of this nature before, and their success with it was less than stellar. I doubt the revolution is going to be nearly as spectacular as Cisco would have us believe.

    HP's UDC (Unified Data Center was first introduced in November of 2001, and I think they perhaps sold six of them. It was a wire-once then run everything virtual solution with whiz-bang GUIs to set up and manage everything, the thing is you had to rip everything you had out and buy all of it from HP. They eventually turned it into a software product that disappeared without a sound.
    IMHO, what killed HP UDC was the high cost to enter and support (and Cisco is well known for their low low prices, right?), which ultimately overshadowed the cost of the labor to do the same without it. I believe that Cisco will have better success than HP or IBM did (mainly because the first step isn't as high), but I really don't see this as a major source of worry for either of them. If HP would just use it's own Procurve group to design their own blade chassis switches they could easily match the Cisco offering; with the UDC software on the storeroom shelves the development time should be short.

    I know you're a Procurve convert Mark, but are you willing to put your foot down in regards to the Proliant folks? If nothing else comes of it, HP should at least be able to produce marketechture in the same league as Cisco.
  • RE: Cisco's big data center plans: Assessing winners and losers rack by rack

    This has failure written all over it. Cisco will have to pour an extreme amount of resources to break in here as the established players; Sun, IBM, HP, and Dell have big infrastructure plays in place. But that's not the real problem: data center servers are a contracting market and will be for several years. Cisco in its impatience to expand, has made mistake after mistake lately (remember the set-top box purchase), but this time it may be a big enough mistake to cause Chambers to take a fall. This is a move that is overreaching and misguided. Ya, failure written all over it!
  • RE: Cisco's big data center plans: Assessing winners and losers rack by rack

    Yes, it is a proprietary solution with just a couple of other players. The ideals of managing the data center infrastructure from a single point is correct, particularly for virtualization, however when this solution doesn't do what you want, you wait.

    Fortunately an open solution has been around since 2003 except in a different area - telecommunications. The AdvancedTCA (Telecommunications Computer Architecture) exists and was developed from the ground up as an industry standard hardware and middleware (management) with more than 100 suppliers and it addresses servers, networking (10GbE fabric), storage and management for health, availability, provisioning and monitoring of just about anything you require including thermal and power usage.

    The Telco's have been doing services (like we will see with cloud computing) for decades and already knew to make the services reliable and affordable you have to do this kind of thinking from the ground up.

    A decade ago they realized you can't keep building proprietary solutions. It's too expensive and the world changes too fast. You have to be agile, and the infrastructure platform has to last much longer with ability to grow.

    If you look you will see the platform already in data centers, albeit for edge applications, but also throughout the network elsewhere. Since 2007 it has become the platform of choice.

    Network Switching, Networking Processing (QoS,Security,WAN acceleration), Blade Servers, and RAID Storage Blades that plug-in anywhere, and cable-less 10 GbE fabric backplane, What are you waiting for?