Takeaway – Corsa Technology is focussed on provided a hardware data plane for SDN WAN use cases.
WAN Networks are different A WAN network has more comprehensive requirements for fabric, shapers and buffers to manage data in a more heavily loaded environment. For example, Broadcom Trident 2 has shallow buffers so that queueing is less effective. For the Data Centre, dropping a few packets at high speed and low latency is perfectly acceptable but in the WAN is not.
SDN Data Plane Corsa Technology has built “forwarding engines” to be a WAN data plane for SDN networking. In SDN you are always buying a data plane to move packets across the network. But can you do this with same silicon everywhere ? For WAN applications you need a flexible silicon fabric that can adapt to the different requirements.
Technology Corsa Technology has built a forwarding engine that uses FPGA technology instead of more common ASIC. FPGA use software programs to implant functions and capability in the silicon plane. The advantage is ultimate flexibility for functions and features at the price of increase software development and moderate cost (ASICs are generally cheaper in volume).
Corsa Technology would propose that FPGA are suitable for SDN WAN application since they can deliver rich functionality and even change the performance of forwarding plane over time.
Today, Corsa is OpenFlow 1.3 compliant and would work with most other OpenFlow controllers.
Importantly, Corsa Technology is shipping 100Gbe networking today.
This hardware is significantly different a typical 1RU whitebox Ethernet switch. The data forwarding capabilities are significantly different and more capable. In the Data Centre traffic is about volume and speed and fibre costs very little.
In the WAN, there is an alternate use case where fibre costs outweigh equipment costs and silicon investments enable better returns.
The EtherealMind View
SDN WAN is the a major trend in the SDN/NFV space and Corsa Technology is well positioned to bring FPGA technology to address some use cases. FPGA have had mixed success in the last 10 years due to software complexity but I believe that new software toolchains and programming practices may see better technology.
I’ll be looking closely at this technology in preparation for a 3hour techtorial at Interop New York on the “10 year strategy” future of the WAN.