My eye-tracking chapter has ended and while I’ve enjoyed in recent months mentoring some very interesting startups that range from hardware acceleration to retail shopping acceleration one phenomenon that has really captured my attention is OpenFlow (www.openflow.org). OpenFlow is actually a specific example of a larger movement called Software Defined Networking. In other words, putting networking functions (from basic routing to security, provisioning, mobility, customization, and management) into easily created and modifiable software that runs in a regular, inexpensive computer and controls regular, inexpensive switching devices simplifies the creation and deployment of new services. The freedom this gives network operators and users – be they public or private – tantalizes the imagination.
What’s especially cool is that new networking functions and operating modes can coexist with current networks on current networks without disrupting existing modes. Thus the exciting possibilities can be explored in an evolutionary fashion, compatible with current equipment. Moreover, because these new methods can be developed on existing networks, not just special-purpose, experimental ones, they can be tried at the scale of today’s real networks. No one can ever afford to build experimental networks of anywhere near the scale of real ones, yet scale matters. So I think we are going to see a blossoming of innovation in networking that is better tailored to applications and users and that the public at large will benefit from.
And now what? OpenFlow!
My eye-tracking chapter has ended and while I’ve enjoyed in recent months mentoring some very interesting startups that range from hardware acceleration to retail shopping acceleration one phenomenon that has really captured my attention is OpenFlow (www.openflow.org). OpenFlow is actually a specific example of a larger movement called Software Defined Networking. In other words, putting networking functions (from basic routing to security, provisioning, mobility, customization, and management) into easily created and modifiable software that runs in a regular, inexpensive computer and controls regular, inexpensive switching devices simplifies the creation and deployment of new services. The freedom this gives network operators and users – be they public or private – tantalizes the imagination.
What’s especially cool is that new networking functions and operating modes can coexist with current networks on current networks without disrupting existing modes. Thus the exciting possibilities can be explored in an evolutionary fashion, compatible with current equipment. Moreover, because these new methods can be developed on existing networks, not just special-purpose, experimental ones, they can be tried at the scale of today’s real networks. No one can ever afford to build experimental networks of anywhere near the scale of real ones, yet scale matters. So I think we are going to see a blossoming of innovation in networking that is better tailored to applications and users and that the public at large will benefit from.