HP beefs up SDN with a new virtual switch

Designed for use with HP FlexFabric 5900AF switch, HP Virtual Switch 5900v provides powerful SDN alternative to anemic VMware vSwitch

Network architectures and SDNs are becoming increasingly crucial components powering the cloud. If an announcement from Hewlett-Packard is any indication, a growing convergence between networking hardware and hypervisor-powered SDN is under way.

Consider HP's newly announced FlexFabric Virtual Switch 5900v software, which is to be used with HP's FlexFabric 5900AF switch. Together, they're designed to replace VMware's Virtual Switch, or vSwitch, and give cloud builders a more powerful way to manage networking in their data centers.

vSwitch was created by VMware to provide a way to manage network traffic among VMs. Unfortunately, vSwitch has lacked many of the more advanced QoS and network management features of its physical brethren, which makes it more difficult to build a robust and scalable network fabric architecture for a cloud. Some have claimed vSwitch isn't really a switch, but more akin to a shared network adapter. It works well for small-scale VM setups but doesn't scale effectively, so it can become a bottleneck.

HP claims Virtual Switch solves this problem by providing each VM with its own dedicated switch port on the 5900AF and by allowing the same level of VM-based management. The Virtual Switch can be used as a drop-in replacement for vSwitch, HP further claims, so one's existing VM configurations don't need to be disrupted to make use of it.

The 5900AF also supports high-end management and flow control features like the virtual Ethernet port aggregation (the 802.1Qbg standard) and the virtual station interface discovery and configuration protocol (VDP). Networking operations that used to have to take place within the VM hypervisor can now be offloaded to the physical switch, says HP.

Another VMware-specific feature for FlexFabric, according to HP, is "the ability to track VM resources during and after vMotion activities." Profiles for such resources will automatically move along with VMs, with no administrative work needed to make it happen.

A need does exist for better network fabric in virtualized environments. Cisco, for instance, has attempted to provide its own software networking appliances to that end. Among them is the Cisco Nexus 1000v, a software appliance that runs Cisco NX-OS and provides policy-controllable software networking for VMware. That said, it's a software-only solution; HP's offering combines VM networking management with an actual hardware solution, although at greater cost.

HP has closely partnered with VMware for some time now, via the former's Converged Infrastructure program. But HP has also long partnered with another company that competes directly with VMware in multiple environments: Microsoft.

Microsoft's Hyper-V virtualization platform has its own software switch -- named, appropriately enough, Virtual Switch -- with a feature set that now compares favorably with VMware's, as well as a third-party hookable API set. (Cisco has since ported the Nexus 1000v to run atop Hyper-V.)

Creating a version of FlexFabric Virtual Switch to augment or replace Hyper-V's Virtual Switch would present some technical challenges, since Hyper-V deals with network adapters a little differently than VMware does. But those issues wouldn't be insurmountable, and those building their clouds with Hyper-V could have another tool in their network-construction kit that wasn't available before either as hardware or software.

The 5900v will go on sale in the fourth quarter of 2013, although HP has not yet set pricing.

This story, "HP beefs up SDN with a new virtual switch," was originally published at InfoWorld.com. Get the first word on what the important tech news really means with the InfoWorld Tech Watch blog. For the latest developments in business technology news, follow InfoWorld.com on Twitter.

Mobile Security Insider: iOS vs. Android vs. BlackBerry vs. Windows Phone
Join the discussion
Be the first to comment on this article. Our Commenting Policies