- Mainstream SDN is ten years out, because enterprise IT does not see a compelling need, buying habits are entrenched and extended product lifetimes are stretching refresh cycles.
- SDN is still nascent technology and has yet to move into the early adopter phase. In fact, from an adoption standpoint, I would say SDN is still in the innovator phase and is thus far from crossing the chasm to early majority.
In my discussions with others (colleagues, vendor employees, other analysts, enterprise IT) about SDN, the discussion usually wanders around to timing. I think it is safe to say that many believe SDN adoption is going to take off – i.e., go mainstream, meaning greater than 50% penetration – in two to three years. When I shake my head and say it’ll be more like ten years, I usually get a look that says, “Hey, Fratto’s gone soft in the head;” then, they point to some corner case as proof, or worse, to someone else who thinks SDN is two to three years out. Here is why I think mainstream SDN is ten years out: lack of a compelling need, entrenched buying habits and extended product lifetimes.
Lack of compelling need is a tough one; I know there is value in SDN technologies, but if I had to justify the hard and soft costs of deploying an SDN compared to the technology in use today, in general, I would have a hard time doing so. Need to integrate your hypervisor with your network? You can do that without SDN. Need to automate changes? You can do that as well. Need to keep virtual servers together? Yep, I can do that too. SDN can help in these cases, but the real power of SDN lies in the future of applications and use cases that have not been thought of or brought into the public eye yet. If and when those applications come along, SDN will follow.
Then, there are buying habits. I cringe a little every time someone says that white box networking will crush commercial networking; such statements ignore how enterprises buy equipment. Small to medium enterprises (SMEs) – anything not a SMB – tend to buy networking from a VAR, and VARs tend to form business relationships with one or two networking vendors, selling their equipment along with design and support services. Unless SMEs and larger companies start buying direct in the next two to three years, or national and international VARs start private labeling white box switches and selling them over their partners, they will continue to buy whatever their VAR offers. Naturally, VARs are in an excellent position to disrupt networking with white box options, because they have that all-important customer relationship built, but such a strategic move would take a great deal of commitment and I do not see it happening.
Finally, there are the extended lifetimes in networking equipment. The rule of thumb is that IT equipment gets replaced every three years, but I suspect for a majority of SMEs on up, the useful lifetime is much longer. In some cases, networking gear simply remains in place. In other cases, it is moved to another role elsewhere. That older gear is working just fine, and unless there is a compelling need to upgrade it, companies will leave it in place.
There has been a lot of good work on SDN technologies within vendors and open communities, and such work will continue, but from an adoption standpoint, SDN is still in the innovator phase and is thus far from crossing the chasm to early majority. I don’t hate SDN and I don’t think it will fail (though its success is not guaranteed and it may go through some more iterations), but enterprise IT moves slowly and ignoring that observation does not make it any less accurate.