Open source SDN let loose, nobody hurt... yet
"Neither," say the companies involved. "It's choice." And 'choice' is probably to be expected at this point in the competition. The five month old multi-vendor OpenDaylight Project, with both IBM and Cisco as active members, has announced it's first open source release code-named Hydrogen.
Hydrogen includes user interfaces, data plane elements and the SDN controller platform. The idea is that the open sourced code will provide an underlying base upon which vendors build their own applications and drive their existing hardware installed base. In effect it provides an agreed 'start line' and a theoretical standard platform pitched so that user organisations (telcos, network operators and enterprises) don't feel locked in and incumbent network infrastructure players don't feel locked out. Pure-play SDN companies naturally worry that OpenDaylight is a disruption disrupter.
Meanwhile Juniper Networks, not content with being a fully paid-up member of OpenDaylight is doing some SDN moonlighting on its own account by open sourcing a version of its just announced Contrail "production-ready SDN solution".
Contrail is Juniper's "network virtualization and intelligence solution" and the company claims it has all the components needed to create a virtual overlay network: SDN controller, vRouter, and analytics engine.
So why the open sourcing of the core code, particularly as Juniper is already involved in OpenDaylight? Juniper reckons it's all part of the SDN love - open sourcing is a way to foster innovation and adoption of SDN. "It's designed to give developers the opportunity to innovate, adopt and experiment with SDN technology that seamlessly integrates with existing network infrastructures," says the press release. The open sourcing should also give potential customers a slightly less 'locked in' feeling.
All this open sourcing generosity naturally has 'pure play' SDN and NFV players, such as Infoblox, watching closely lest they get elbowed out of the way by the incumbent suppliers.
That uncomfortable bulldozed feeling had already been visited upon SDN pioneer Big Switch, which pulled out of OpenDaylight alleging that big vendor politics was already at work and had overseen what it regarded as inferior controller technology from Cisco being favoured above its own (see - SDN strife: Big Switch pulls back from OpenDaylight)
Such spats are to be expected of course, but now that Hydrogen, in particular, is out of the box, what did another 'pure play' network software and automation provider think it indicated.
According to Stu Bailey, CTO and Founder of Infoblox, anything that helps the network market move away from the dedicated-function boxes and custom ASICs of the current hardware defined network paradigm is alright with him.
"To the extent OpenDaylight is freely available and promotes an SDN mindset, it's good," he says. But there are niggles.
"While they mention them in their release, OpenDaylight has been slow to focus on network whiteboxes and OpenFlow-enabled devices. For instance, the announcement mentions OpenFlow 1.3.0. Why not the latest version, OpenFlow 1.3.2, which was released in April of 2013? Those kinds of choices raise questions."
For Bailey nearly the whole point of SDN and NFV is to assist the migration to what he calls 'White Boxes'.
"If OpenDaylight is truly focused on enabling businesses to do more with less in the networking space you'd see a rigorous focus on network whiteboxes, OpenFlow enabled devices, and Open vSwitches, since that is where the economic gains exist to fund SDN on a massive scale. Also the focus should be on a truly scalable distributed control plane with SDN compiler, language, and debugging support... Nothing from the OpenDaylight Hydrogen release announcement suggests that is their direction."
So while OpenDaylight is helping scales fall from eyes, it's not likely in itself to be the destination, more a useful starting-point... at least according to Bailey. In the end he believes the best software will win out.
"In the short term, the introduction of OpenDaylight has slowed down SDN adoption and caused people to be tentative about whether incumbents will dominate the near term business opportunities. That has not been helpful to the buyers of networking solutions. But in the end the evolution to software is inevitable. The economics of inexpensive programmable Ethernet and SDN are so compelling that no one vendor's efforts will be able to contain the paradigm shift."