Enterprise packet switching hardware has continued to be a hold-out against open source in the data center. The reasons for this include many of the same, tired, old arguments used in earlier fights around open source. But, at least part of the issue has been that such devices have traditionally included an under-powered CPU complex where modern Linux was just not practical anyway.
For a variety of reasons, designs are appearing that couple modern packet switching hardware with CPUs comparable to those used in contemporary server machines. Couple that with efforts like the Open Compute Networking Project, and the possibilities for open source networking infrastructure in the datacenter starts to look like a real possibility.
How would such a box look? What would we do with it? What should we do to get it? These are the questions this talk intends to address.
Survey this Session