3 Comments

Summary:

At its Cisco Live event today in Las Vegas, Cisco rolled out a range of new features and products designed to convince companies that Cisco is all they need for their cloud network and server needs. But will customers actually buy into Cisco’s vision?

Lew Tucker - VP and CTO, Cloud Computing, Cisco - Structure 2011

Lew Tucker - VP and CTO, Cloud Computing, Cisco - Structure 2011At its Cisco Live event today in Las Vegas, Cisco rolled out a range of new features and products designed to convince companies that Cisco is all they need for their cloud network and server needs. Seriously — its new products range from fabric extenders for its Unified Computing System server gear to cloud-based email filtering.

But is Cisco really cut out to contend from the data center up through cloud services?

In the data center, at least, Cisco looks really good. Servers are its brightest star right now, and Cisco VP Soni Jiandani shared the statistics: roughly 5,400 customers as of its most-recent earnings statement, adding about 1,000 per quarter, and presently No. 3 in blade market share according to IDC at nearly 10 percent. In the United States, IDC says Cisco is the No. 2 provider behind only HP, commanding 19.7 percent of the market. While other business units are struggling to maintain, servers have grown into nearly a billion-dollar business in less than two years.

Cisco’s news today was mostly technological and not strategic, but in light of rumors of massive layoffs and CEO John Chambers acknowledging a “leaner” Cisco during his morning keynote, every little thing matters. If Cisco does intend to lean on UCS as a future revenue driver, how that gear performs could be just as important as Cisco’s overall data center and cloud computing strategies.

Most of Cisco’s news today was new features designed to make its UCS gear even more appealing by letting customers do a lot more with them. That means lots more bandwidth to run lots more workloads simultaneously, and unified storage management regardless the connection (Ethernet, Fibre Channel or Fibre Channel over Ethernet).

What relates this all to cloud computing, at least in terms of private clouds, is network virtualization. Much of the new UCS fabric capabilities are carried out virtually, meaning bandwidth, connections and other changes happen in the software, as opposed to in the hardware. UCS-based networks can flex and bend as needed without miles of cable and complex configurations.

Cisco also touted its new service profiles feature, which looks to be the result of its LineSider acquisition last year. Essentially, policies around security, bandwidth requirements, storage connections, etc., stay with a virtual machine wherever it happens to move across the network. This means workloads can move dynamically as might be required within a private cloud, and they automatically maintain those policies as they migrate to new physical resources.

When viewed in light of its purchase of private-cloud, self-service software from newScale and its support of OpenStack at the infrastructure orchestration layer, one starts to get the impression that Cisco is putting together a solid lineup of technology for managing and delivering virtual infrastructure.

Given its recent failures in the consumer application and corporate collaboration spaces (Cisco recently killed its cloud-based email service), Cisco’s new IronPort email security and WAAS WAN-optimization products also seem pretty smart. By using network intelligence to improve the security and performance of cloud applications rather than trying to deliver applications itself, Cisco is operating much more within its wheelhouse.

But Cisco’s biggest problem isn’t capabilities as much as it is price and proprietary software, especially in the data center. To get the performance that Cisco’s offering, you have to buy into its soup-to-nuts suite of gear. That means servers, switches, software, you name it. That might not be a problem for large businesses willing to pay top-dollar for performance and risk aversion, but there aren’t too many of those, relatively speaking.

Everyone else might be looking to meet their cloud needs by purchasing less-expensive, more-commodity gear from a vendor like Dell, or by cobbling together best-of-breed infrastructures using different vendors at every layer. Presumably, many more will soon start looking at options like OpenFlow-compatible gear, which lets users manage their networks on their own terms. Cisco doesn’t seem to have answered these issues just yet.

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

  1. Derrick:

    Thanks for the thorough write-up on this weeks’ announcements. So, you bring up some good questions and I wanted to explore your points around our solutions being proprietary.

    While we expend a great deal of effort baking innovation into out products, we also spend a fair bit of time ensuring standards compliance. Cisco’s data center solutions are not dependent on there being a pure Cisco stack from top to bottom. If you use UCS as an example, at the end of the day, the UCS connects into the rest of the data center through fully standards compliant protocols like Ethernet and Fibre Channel and will happily work with anyone’s switches. The UCS is managed via traditional management protocols as the large management ecosystem for UCS will attest. As another example, the Nexus 1000V, our soft switch for vSphere, runs on anyone’s compute hardware and will connect to anyone’s switches.

    On the open source side of things, we are actively participating in OpenStack. Cisco is also an member of the Open Networking Foundation and are exploring how to bring OpenFlow-based products to market. So, stay tuned fro more developments on both those fronts.

    Regards,

    Omar

    Omar Sultan (@omarsultan)
    Cisco

    1. Omar,

      Thanks for the reply and for clarifying Cisco’s involvement in standards and open source. I’m particularly interested in seeing how OpenStack and Open Flow involvement play out on the product front.

      Regarding standards, I think the issue around any company — not just Cisco — selling a top to bottom line of products is that they’re generally optimized to work together. While they technically can work with other vendors’ products, the advertised features, specs, performance metrics, etc., are for single-vendors deployments. In Cisco’s case, UCS, Nexus, etc., are designed with each other in mind, no?

      Not that it’s not the way it should be, just that it tends to encourage a single-vendor approach, which might be more expensive than best-of-breed products. If I’m wrong, though, please do correct me.

  2. Hi Derrick,

    Upon reading this I felt compelled to respond for the same reason that Omar offered his opinions. Omar addressed the openess issue, so I won’t dwell on it except to say that UCS in no way dictates the selection of a core data center switch in the same manner that a Nexus network doesn’t dictate the choice of a server vendor. If our customers see Opex and Capex benefits from a Cisco network and compute (cloud) infrastructure that’s a good thing, but the choice is theirs.

    Also, there is a factual inaccuracy in your piece. Service Profiles and their ability to abstract the workload from the underlying hardware have been integral to the UCS design from the very beginning (and they shipped with the platform ~2 years ago). While a fantastic innovation in their own right — and a real point of differentiation for UCS — Service Profiles are unrelated to the LineSider acquisition.

    Thanks —

    Jesse Freund
    Cisco

Comments have been disabled for this post