How does a private cloud architecture change your server I/O? This is an important question if you’re looking at moving to a fully virtualized data center. Because the decisions you make today about how to connect your servers will directly impact your application performance and your ability to fully leverage your assets down the road.

Take a quick look at this video. Then read below for two new features from Xsigo that deliver scalable management and 4X the performance of 10G Ethernet.

Cloud deployments up the ante for server I/O in at least two ways.

Cloud Requirement: More I/O Bandwidth

Server bandwidth demands have changed… a lot. When Xsigo introduced 10Gb server connections three years ago, our customers though that was plenty. And it was… for the servers and software they had then. A year later, Xsigo’s launch of 20Gb connections again provided ample headroom. But now customers tell us they want more. And there are at least three good reasons for this.

1) IT managers now want to virtualize more applications, including Exchange, and database. These I/O-intensive applications, previously considered taboo for virtualization, are more and more becoming fair game.

2) There are now more VMs per server. It used to be “five VMs per server,” then ten. Now 20 VMs per server is pretty common. This increased loading directly increases I/O requirements.

3) The hardware and software are more capable. Nehalem and vSphere tripled your I/O capacity vs. the previous generation products. That’s huge. If you’re still operating with the same old pipes out of the server, that may be your bottleneck.

The bottom line is that bandwidth needs are much higher today.

NEW 40Gb Server Links

Xsigo is responding with new 40Gb server links. That’s a lot of bandwidth, the equivalent of 15,000 simultaneous Netflix HD video streams. But it’s needed if you’re planning on running I/O intensive apps.

Cloud Requirement: Fluid I/O management

The second fundamental change is that I/O must be managed in a way that is fluid.

Face it, managing every port and NIC and HBA on a one-by-one basis (and dealing with the cables in between) does not foster agility.

If you want to change something on 100 servers, how do you accomplish that quickly? If you want to deploy a new network (to support a new VMware feature such as fault tolerance, for example), how do you do that? With traditional I/O, you’d probably rather deploy a new server than take a production server down to install new cards and cables.

The cloud has to be different. I/O must be re-configurable when needs change.

NEW XMS 3.0 Management Interface

Xsigo’s new management interface introduces several important new capabilities that address the needs of the cloud:

• User-defined I/O templates: rollout or modify connectivity on thousands of servers in seconds
• Unified connectivity management: deploy, view and manage connectivity on thousands of servers from a single pane of glass
• Performance monitoring: monitor and manage I/O performance across all servers to quickly identify and remedy I/O bottlenecks
• Simplified disaster recovery: wide-area I/O management lets users manage and migrate I/O across sites to simplify disaster recovery, move entire I/O identities from one site to another, and recreate an entire configuration at a remote site so applications can be back up and running in minutes
• iPad accessable: the new management interface can be accessed through an iPad. With the touch of a finger, users can create and manage data center server I/O using drag-and-drop features.

The combination changes I/O management in ways that would not have been possible before virtual I/O.

But that’s the point. You don’t virtualize to do things the old way. You virtualize to achieve fundamental change that makes the data center and the people that run it all more efficient and productive. That’s the whole idea of the private cloud, right? Now that fundamental change — and the private cloud — has come to I/O.