37 Comments

Summary:

Today Cisco announced its much awaited data center play with what it calls its Unified Computing System. Om does a great job explaining why the networking giant is moving into the data center as the demands of digital data tax the current three-part IT infrastructure of […]

Today Cisco announced its much awaited data center play with what it calls its Unified Computing System. Om does a great job explaining why the networking giant is moving into the data center as the demands of digital data tax the current three-part IT infrastructure of servers, networks and storage.

The current infrastructure is showing its age, so three years ago Cisco set out to address the problems with its Unified Computing System. The end device is a reinvention of the server by looking at it as part of the network fabric and not as a discrete element attached to the network.

Hewlett-Packard and IBM will bear the brunt of Cisco’s push into the data center, while companies that have a healthier commodity and lower-end server business such as Dell will likely feel less pain in the near term because some of their customers will not need the God-like powers Cisco is promising with its newest system.

Om laid out some speculative system specs, but Cisco didn’t give out anything related to the processors or the memory available on the system, saying those details will emerge later this month from Intel. A spokeswoman did confirm that the system will use the latest Intel processors, which have a technology used to speed up access to memory at the chip level as a way to reduce computation bottlenecks associated with multicore chips.

So what do we know?

  • Each chassis could have up to eight blades with embedded “integrated fabric interconnect”
  • The blades can have a 20-port or 40-port switch for 10 Gigabit Ethernet
  • The switches have two converged network adapters that can support Ethernet or Fibre Channel
  • The fabric interconnect can connect up to 40 chassis for up to 320 blades managed as a single system
  • The system is managed through two pieces of software: one called VMLink that was developed in conjunction with VMware that allocates network resources to each virtual machines and another that is a management console and server provisioning tool for the whole system.
  • Pricing won’t be available until the system is generally available sometime in the second quarter of 2009.

The Unified Computing System is an entirely new class of data center equipment, designed specifically for managing thousands of virtual machines on one system (although it can also handle non virtualized workloads as well.) Cisco will also release open APIs for the system, which will allow it to work with legacy data center equipment.

Agreements with partners have resulted in software for the new box that allows for a virtual machine to be instantly provisioned on whatever operating system the customer would like. One notable missing partner was Citrix, which owns the open source Xen hypervisor used in many of today’s largest cloud computing installations. Cisco says it is still working to bring  partners into the ecosystem, and that is also trying to get VMLink set as an open standard, so other hypervisor vendors could build that into their products.

Xen may not be supported, but Cisco has worked with myriad partners for this system, which include vendors from Red Hat to EMC. One of the clearest beneficiaries of the Cisco push will be BMC Software, which makes IT management software that has been overshadowed by offerings from HP and IBM. The new competition between the larger IT management software providers and Cisco may give BMC the opportunity to push its way into more customer sites if Cisco’s system gets adopted.

Cisco has 10 beta customer sites currently using the system, including Savvis, and it plans to go after enterprise customers. That will require Cisco to shift its sales strategy; instead of going after its traditional relationships on the networking side, it will instead have to sell to the CIO.

To that end, Cisco is advertising a 50 percent reduction in support infrastructure such as switches and cabling associated with each server, and up to 35 percent reduction in capital expenditures and operational expenditures associated with buying data center gear, if a company switches to its Unified Computing System.

Right now, the Cisco focus appears to be on what this new hardware can do for virtualization and web scale computing — be it a Facebook’s infrastructure or a cloud provider’s. In a call this morning, John Chambers, Cisco’s CEO, also focused on what it could do for the bottom line, saying it gives Cisco access to about a quarter of the many billions spent inside the data center, up from less than 10 percent presently. That potential revenue is why Cisco is reinventing the basic building block of the data center.

You’re subscribed! If you like, you can update your settings

  1. Adrian Ionel Monday, March 16, 2009

    They enter a crowded market going against folks like HP and IBM. At the very high-end CSCO lacks the products and credibility and in the mid-range it will take time to build scale. Even then it will be low-margin dog fight. Virtualization/consolidation is huge but they don’t seem to have anything proprietary there. Is this a big opportunity for them or just a defensive move?

  2. Stacey:

    Your comment that the Cisco UCS is a new class of data center equipment is spot on. From our perspective, this is a perfect example of “skating to where the puck will be.” As Chambers said this morning, we are not responding to competitors, we are responding to market transitions. Data center virtualization is a foregone conclusion and the Cisco UCS was designed for that environment. In fact, by taming the operational complexity of virtualization we feel we can help customers accelerate the rate at which they can deploy virtualization and its attendant benefits.

    Omar Sultan
    Cisco

  3. Cisco’s Data Center Moves: Who Wins, Who Loses? Monday, March 16, 2009

    [...] layer (thanks to virtualization technologies) that is managed by a specialized piece of software. Stacey has captured the intricate details of the news, while I have already posted about the imperative behind these [...]

  4. Omar,

    How does this solution address disaster recovery and storage utilization that point SAN solutions do not address more comprehensively? SAN offers many benefits such as block replication, which facilitates DR plans, Snapshots, LUN resizing, thin provisioning, clones and the benefit of speed writing to many disks to name a few. For my environment, having a true common pool of storage and its software features enable us.

    How is abandoning that solution better for us in the long term? A 48 gig front end server for VMware is $15K. Storage in the SAN is another $20K. Switching for 96 ports of ISCSI Ethernet another $15-25K with aggressive discounts at your primary competitor. Your solution is going to cost less minus software costs, and provide more functionality? I don’t buy it.

    The blind grab at market share and clout wont work. As the previous reader stated, you have little credibility. Your competitors are cheaper, and have faster gear. EIGRP was good for vendor lock-in. Engineers know better now.

  5. Cisco’s Data Center Moves: Who Wins, Who Loses? « Ads Sanrego Tuesday, March 17, 2009

    [...] layer (thanks to virtualization technologies) that is managed by a specialized piece of software. Stacey has captured the intricate details of the news, while I have already posted about the imperative behind these [...]

  6. Jesse Kopelman Tuesday, March 17, 2009

    Will IBM bear the brunt or will revenues just shift from one business unit to another? Various IBM Consulting Units make tons of money selling solutions that revolve around Cisco gear.

    1. Exactly. IBM consulting is bigger than IBM hardware. There is no reason why every TJMax needs an overpriced Cisco switch when a just as overpriced and maybe fatter commission Nortel will do just fine. That sharp dressed guy at IBM will convince them it is better faster more reliable when the memo goes out, “do not sell Cisco, they are our competitor”.

      Although I think the superfast backplane that can provision 10GE or FC on each port is pretty cool :)

  7. Unifying the Data Center/Cloud Infrastructure « Yes Cloud! Tuesday, March 17, 2009

    [...] GigaOm provided some details on the announced products. How are these products different that what is available today? Few immediate thoughts: [...]

  8. Prashant Gandhi Tuesday, March 17, 2009

    Overall, this is a cool architectural-centric product offering for next generation data center and cloud computing infrastructure, consisting of an end-to-end centrally managed solution. No doubt Cisco has up the ante in the data center. It’ll be interesting to see how other data center vendors respond – via their own product innovations, M&A activities and/or partnership re-alignments.

    Your product summary led me to dig deeper on the technical details of UCS portfolio; see http://yescloud.wordpress.com.

    PG.

  9. Sun, IBM, and the value of a comprehensive proposition | Paul Miller Wednesday, March 18, 2009

    [...] Cisco’s Data Center Play Reinvents The Server (gigaom.com) [...]

  10. Robin Harris of StorageMojo his the nail on the head re: Cisco « Yet Another VC Blog Friday, March 20, 2009

    [...] Cisco’s Data Center Play Reinvents The Server (gigaom.com) [...]

Comments have been disabled for this post