It was a dark stormy autumn night and three networking engineers had nothing better to do than ponder the heavy topics of transactional consistency in a distributed SDN environment in Episode 16 of Software Gone Wild podcast.
Here are a few of the topics that crossed our minds:
- What happens when an SDN-driven network update fails?
- Will SDN save us from manual errors? Will it make us more precise?
- How will we cope with API/CLI changes between software releases?
- Automating everything might result in making a big mess out of everything.
- We should design for failures, not try to do everything to avoid them, and continuously test our network under failure scenarios
- Should we leave the underlay alone and never touch it (like we never touched the optical transport two decades ago)?
- We have to change our way of thinking and get used (again) to the benefits of decoupling;
Finally, there were two evergreen topics we couldn’t possibly avoid:
- Per-application appliances as compared to big bad boxes;
- Lack of ASICs in network appliances.
And here’s the list of things we mentioned (and promised to get them into show notes):
- NTT Global IP Network Configuration Tool (NANOG 54): http://www.nanog.org/meetings/nanog54/presentations/Tuesday/Morris.pdf
- Antifragile organization: http://queue.acm.org/detail.cfm?id=2499552
- Resilience engineering - learn to embrace failure: http://queue.acm.org/detail.cfm?id=2371297