“I’m not dead yet” and other observations on ONS 2017
So after attending ONS 2017 and listening to the various keynotes and sessions, I came away with a few impressions, which I list below in no particular order. Apologies to Monty Python and any others whom I may have plagiarized or insulted.
“I’m not dead yet, part 1” — ONS
Just when some thought that ONS was being marginalized or shunted off to the side in the battle for SDN mindshare, along comes this conference, with presentations by Martin Casado, Google, AT&T, Intel, Huawei, and others. Former ODL grand poohbah Neela Jacques, former ONF grand poohbah Dan Pitt were in attendance, along with technical top dogs from Cisco, Brocade (speaking of “I’m not dead yet”), Linux Foundation, Red Hat, ONOS, and others. A very good turnout in that so much of the SDN-osphere were present.
So rumors of ONS’s demise seem also to have been greatly exaggerated.
“It’s just a flesh wound” — OpenDaylight
Some were wondering about the future of OpenDaylight, given the following:
- Cisco end-of-life for their ODL-based controller
- Brocade (with their own ODL-based controller) gets acquired by Broadcomm, leaving their SDN team in a bit of limbo
- Ericsson (also with their own ODL-based controller) having bottom-line issues and undergoing layoffs
- No ODL Summit planned 2017
But in spite of this, ODL developers and promoters were present in force at ONS, possibly in part because the Linux Foundation is running both ODL and ONOS. There were sessions on ODL, various keynotes mentioned deployments using ODL, etc.
So all-in-all, ODL seems to be in much better shape than Monty Python’s Black Knight.
That’s good news to SDN proponents. Of course, with the ever-widening definition (or definitions) of SDN, it is easy to understand how everybody is doing it. However, the good news, as evidenced by the turnout at ONS, is that vendors and customers alike are implementing some form of centralized, programmable, automated networks. Which is a break from our static, rigid, and inflexible past.
“I’m not dead yet, part 2” — NFV
There was a session discussing why NFV has failed, and how it can be fixed. That session promoted the idea that network functions should not be deployed as VMs or containers, but rather as applications. The premise being that VMs/Containers are too heavy and costly and difficult to manage, while we have a large body of successful application deployment solutions, and NFV would be wise to follow suit.
Not everybody agrees that NFV is unsuccessful, as evidenced by the significant number of sessions and keynotes highlighting progress towards successful NFV deployments. But the argument regarding implementing virtual network functions as applications is interesting and bears consideration.
The comment “Rumors of my demise have been greatly exaggerated”, or something similar, has been attributed to Mark Twain, but probably could be similarly applied to SDN and NFV at this moment in time, and to the Open Networking Summit in particular. Here’s to hoping the Linux Foundation continues to provide such a forum for discussion and sharing of all types of SDN and NFV ideas in the future.