When OpenDaylight was announced, we at Gartner were skeptical because the consortium was led and comprised of the vendor/selling community with near-zero participation from the user/buying community. Thus, our concerns were that it didn’t have the best interests of enterprise customers in mind. It had the tell-tale signs of a delay tactic to slow SDN deployments, while giving incumbents a chance to “catch-up” to non-incumbents and startups. This week, OpenDaylight announced their first code release (Hydrogen) so it is a good time to revisit this discussion. Let’s start by looking at the facts surrounding what has happened since OpenDaylight was commissioned/announced (in April 2013) and today…
- In November, Cisco (a founding member of OpenDaylight) announced their next-gen strategic data center platform, ACI.
- Several other vendors including Juniper and VMware released mainstream SDN solutions including NSX and Contrail, respectively.
- Juniper even open-sourced a version of their Contrail controller, which now competes with OpenDaylight.
- SDN pioneer Big Switch reduced their participation in OpenDaylight and ultimately shifted corporate focus.
- Many commercial SDN offerings do not support or integrate with OpenDaylight (i.e., Juniper, Dell, HP and VMware). Cisco supports Daylight in some products, but not in their flagship data center network solution, ACI (at least not yet).
- Many additional member organizations were added to the OpenDaylight effort, but the buying/user community remains mostly absent from the membership base (unlike the ONF).
- While it isn’t anywhere near a consensus among Gartner clients, several have remarked to me that they wanted to wait for OpenDaylight before making any “major decisions” around SDN.
So overall, this seems to confirm our initial concerns with the effort in many regards.
However, at this point, that is essentially water under the bridge and we can look at OpenDaylight moving forward… So there is now usable code, derived from a wide variety of contributors and based on what I’ve read and heard, it includes some good stuff. How does it compare with or differentiate from other open source controllers, is it reliable and does it scale? Short answer is we don’t know; it is too early to say.
However, OpenDaylight is still very much a science project for mainstream enterprise. Most mainstream network designers want a supported enterprise-class product which typically means an appliance, and maintenance/ support for it. OpenDaylight simply isn’t there yet, and outside of IBM’s SDN offering, there isn’t widespread commercialized product support for it either in the data center (at least not yet). Time will tell and perhaps more mainstream networking vendors will integrate it into their SDN solutions and/or wrap sheet metal around it and sell it with a support contract. But at this point in time, it competes with other open-source controller solutions such as OpenContrail and Floodlight (and others). And to date, we’ve seen very limited adoption of these in mainstream enterprise, and don’t anticipate that changing much over the next year or two.
The Bottom Line
At this point in time, I am not saying that OpenDaylight is a bad thing, in fact quite the contrary – Open Source SDN controllers are a very good thing – they foster SDN adoption/innovation and lead to increased customer choice. But you have to take into account the full backstory of OpenDaylight, including what has been left in its wake.
Thanks for reading and looking forward to your thoughts and feedback…
PS-Some Published Research regarding OpenDaylight, NSX and ACI