A third version of the OpenDaylight Project's software-defined networking (SDN) system, codenamed Lithium, is now in general release. With it comes a better sense of who OpenDaylight is appealing to most strongly: not enterprises, but rather the same telecoms who have also gravitated to OpenStack to build their new infrastructures.
The real appeal of OpenDaylight
OpenDaylight's main selling point is that it provides both SDN and network function virtualization in an open source package. OpenStack, in turn, offers a Neutron networking-as-a-service layer, a way to build cloud-based network topologies. Some of the most prominently discussed new features in Lithium integrate directly with Neutron -- e.g., virtual tenant networking, a way to map multiple virtual network designs onto most any physical network layout.
The advantages of such a technology for telecoms are clear, along with the crop of automation functions rolled into Lithium -- e.g., the Time Series Data Repository (TSDR) feature, which can harvest network activity data and make it available for analysis, or the Topology Processing Framework, which is designed for viewing a network via various filters or aggregations.
For more evidence of how telecom-centric OpenDaylight is turning out to be, look no further than the use case profiles on the OpenDaylight Foundation's blog. Most of the outfits discussed there are telecoms -- Orange, AT&T, Comcast -- or affiliates of the telecom industry, such as CableLabs. A recently formed advisory group is also chaired mostly by folks from the telecom side, with a sprinkling of academics (and one seat occupied by a representative of NASDAQ).
Not the only open door in town
So far OpenDaylight has been as notable for the politics surrounding the consortium as it has for the software itself. It's been branded as being more a vendor consortium (with Cisco and IBM in the lead) than a user-driven community effort, as opposed to projects like OpenFlow. Having the Linux Foundation as the umbrella for OpenDaylight hasn't quelled skepticism about the project's goals, either.
If OpenDaylight plans a bid for legitimacy with enterprises by showing that its solution is carrier-grade, the effort may backfire -- not least because carriers themselves are keeping their options open.
Aside from the existing OpenFlow standard, there's the Open Networking Lab, a project from Stanford University and UC Berkeley and with contributions from a number of carriers -- including AT&T. That project's OpenDaylight competitor, the Open Network Operating System, surfaced late last year and is intended to be of more direct value to carriers and network builders than to hardware vendors. In the long run it may appeal to mainstream enterprises as well -- assuming it doesn't bring the same tedious complexity that has made projects like OpenStack unappealing.