BigSwitch Networks, a Palo Alto, Calif.-based network virtualization startup founded on the principles of the OpenFlow standard, has raised $13.75 million in a Series A financing, led by Index Ventures and Khosla Ventures. The company’s stated goal is to become the VMware of networking by utilizing the newfound ability to program enterprise networking gear, a capability made possible by OpenFlow.
BigSwitch might just have the open networking chops to pull off its mission. The company has joined the Open Networking Foundation, an organization committed to pushing the OpenFlow standard and other software-defined networking protocols, and which is backed by major web companies and telcos. Oh, and Co-Founder and CEO Guido Appenzeller was part of the Stanford University team that developed the OpenFlow v1.0 standard.
Much like server virtualization gives IT administrators the flexibility of creating virtual server infrastructures not constrained by the limitations of their physical hosts, network virtualization lets network administrators create virtual network topologies that make managing sprawling IT networks a far easier task. This is because network administrators can assign network devices the resources they require without worrying undertaking the legwork of creating actual physical connections.
It’s a networking model that already has its champions, including Xsigo Systems with its virtual I/O products and Cisco, which does LAN virtualization. BigSwitch says its name is derived from the concept of managing the network like “one big switch,” which actually is the goal behind Juniper Networks’ recently announced QFabric architecture, too.
However, these vendors generally are trying to accomplish network virtualization by adding special physical appliances or requiring new, likely expensive switches. By leveraging the OpenFlow standard, BigSwitch hopes to alleviate the need for new gear by providing the virtualization capabilities on commodity gear.
As BigSwitch describes its ambitions on its website:
Prior to [OpenFlow], we’ve simply never had a programmatic way to reach deep enough in to enterprise-grade networking devices to do what we need to do. Configuring a network, to us, is modifying the inputs used in network algorithms imprinted deep in embedded hardware and software. Programming a network, which is what we need to do, implies relying on these algorithms most of the time but over-riding them every once in a while. A simple case in point: our virtual networks don’t need spanning tree.
Or, as my colleague Stacey Higginbotham has summarized OpenFlow:
The idea behind the OpenFlow effort is that today’s network needs to be smarter and more flexible in order to handle and efficiently deliver more information. To do that, the fundamental idea is to separate the packet switching mechanisms and control functions. Users can freely develop and operate control middleware independently of the switching mechanism.