In part 9 of this series, I considered whether SDN was mature enough for most companies to benefit from it. I’m going to wrap up part 10 of the series by reflecting on the key question I put forth at the beginning:
Should an organization embrace SDN now, soon, or hold off for now?
Most organizations are keen to learn about SDN but are not yet ready to deploy in production. Many are experimenting with various SDN technologies to try to assess which vendor’s solution is best from them – be it Cisco’s ACI (all encompassing, but expensive), VMware’s NSX / Juniper Contrail (cheaper, simpler, tunneling technology but with concerns around the lack of application path visibility) or an open solution such as OpenFlow which avoids vendor lock in, offers the best chance of API standardization and therefore should provide a wider range of orchestration application options, but without enterprise-grade support.
With the lack of compelling business drivers, most organizations are leaving deployment to those who are less risk averse, can justify the expense of being in the vanguard for a specific SDN benefit, or are more experimental in nature (e.g., universities).
My advice for now is to watch, read, experiment, learn from others’ mistakes and be ready to engage when the appropriate business drivers emerge and the technology stabilizes and is proven. Also assess NMS vendor’s current offerings (and roadmaps) in light of the issues discussed above. The only exception to waiting is where a significant network upgrade is imminent in which case consideration should be made of the possibility of purchasing SDN compliant hardware (e.g., networking infrastructure with OpenFlow support, ACI supported hardware, etc) as a possible precursor to future SDN deployment.
To return to the beginning of my series on SDN, click here.