Ethernet switching is in a sea change, and server virtualization is a big part of the reason why. Gartner recently published a new competitive landscape for data center Ethernet switches. In that report, titled “Competitive Landscape: Data Center Ethernet Switches, Worldwide, 2011,” Gartner made several interesting comments about this evolving space and how trends such as virtualization and cloud computing change data center Ethernet switch requirements.
This report includes Xsigo in the data center Ethernet switch category, adding Xsigo to the “Data Center Network Specialist” group, along with Brocade, Extreme Networks, and Juniper Systems.
The definition of this group is vendors that “depend on their dedicated focus and specialization in networking to innovate and meet the fast-evolving needs of customers.”
The words “fast evolving” nicely sum up what we see happening.
In the past, there was no clear distinction between the switches found in a data center and those used elsewhere in the organization.
But this is now changing. Virtualization changes the game, not just for the way we run applications on servers, but also for the way those applications interact with all of the other data center resources.
For that reason, the data center Ethernet switch is taking on a new look. Gartner characterized this ongoing transition as “an important step in the evolution from box-centric, data-center-optimized Ethernet solutions to a fabric-based approach for large data center and cloud infrastructure.”
In one sense, you could say that actually very little has changed. Applications still need access to specific resources: networks, storage, and the devices that surround them, such as load balancers, security devices, and other appliances. And the applications still require sufficient bandwidth to ensure that performance meets the objectives.
The difference with virtualization is this: those exact same requirements must now be met in a dynamic world, where multiple applications could reside on any server, not just one application on a dedicated device. This is a big change.
Here are three examples of the issues this creates for the switching infrastructure:
- The “East-West” problem: In the old days, most data centers were laid out in nice, orderly tiers, allowing data to flow in predictable paths. But with virtualization, applications can run anywhere. This creates messy server-to-server data paths where data must negotiate multiple layers of infrastructure to reach its destination. The resulting latencies, bottlenecks and general bad behavior are called the “East-West Problem.” If we’re going virtualize more, and make data centers truly resemble a cloud, this must be addressed.
- Exponential performance demands: Virtualized servers demand a lot more I/O bandwidth than bare metal servers do. At Xsigo, we’ve seen this firsthand. Over the past four years we have upgraded the performance of our server interconnect from 10Gb, to 20Gb, and then to 40Gb. While the rest of the industry has experienced a somewhat leisurely transition from 1Gb to 10Gb connections, our customers (most of whom are hard-core virtualization users) have hungrily jumped to the highest available speed. They have simply found that virtualization works better with more bandwidth.
- VLAN Exhaustion: VLANs did a great job of addressing the problem they were created to solve. But when it comes to creating large numbers of isolated interconnects within an agile data center, they don’t scale well. You can run out of VLANs. Or they can become a management nightmare to administer. Eliminating this problem — while providing the needed isolated connections — is a challenge that must be overcome.
Gartner stated in the report that, “data center switching is a fast-evolving market with a different problem set and several new technologies likely to disrupt legacies.”
At Xsigo, we see this happening already.
The new Xsigo Server Fabric addresses the needs of scalable, high-performance Ethernet switching with virtual I/O capabilities that are ideally suited to the needs of a dynamic data center.
The Xsigo approach is different from the traditional network switch, yet it’s being deployed in support of very traditional, mainstream applications within large organizations.
And in that way, it meets the most important requirement of all: to provide a seamless, non-disruptive path forward towards a more efficient, lower-cost data center.
If you’d like a copy of this report, please email me at jtoor “at” xsigo.com.
Or Gartner clients can access this report here.
