While I know and respect Joe, in this case I believe he is only partly right. He states that SDN is an architecture, not a market. He correctly points out that technologies such as OpenFlow will be embedded in switches and routers and do not themselves make a market. In this case the market remains the switch and router market. True. But his is a narrow view in my opinion for two reasons:
First: As Tom Nolle points out in his blog, IP is an architecture, but it’s also a market. Same goes for cloud computing. As SDN takes off, more and more SDN-enabled products will emerge and it will become necessary to separate them, for market sizing and many other reasons, from legacy products.
Second: SDN opens up new opportunities for agile service delivery but will also create the need for new products. For example, a regional service provider who offers enterprise VPN and consumer Internet services sees very different usage profiles during the evenings, when consumer consumption of OTT services ramps up, vs. during the day. The two also have quite different quality of service requirements as well as consequences for service outages or poor performance. Adapting to these different demands by being able to [re-]provision network resources dynamically through centralized software control – the promise of SDN – can offer significant Capex and Opex savings and help create competitive differentiation.
But none of this will happen without a new breed of analytics and orchestration products. And if you believe the network infrastructure suppliers will include these capabilities as features in their products (an existing market), history tells us otherwise. A lot of “SDN-ready” announcements are marketing hype and frankly, infrastructure vendors in general have a poor track record of developing best-in-class management tools. Today’s $1.2 billion (according to Gartner) network performance management & diagnostics market suggests that there will be plenty of need/opportunity for independent management
solutions. Packet Design’s own research bears this out. Virtually every service provider surveyed said that SDN requires new management tools and that some of their existing tools will not work with SDN.
I do agree with Joe that there’s a lot of hooey when it comes to SDN market predictions. I’ve seen reports forecasting the SDN market to reach $35 billion by 2020. As enthusiastic about SDN as we are at Packet Design, we recognize that it will be years in coming, and we strongly advocate and believe that SDN will be an evolution not a revolution (as forecasts like this suggest).
Forklift upgrades are not feasible – the returns just won’t be there. Most forecasting models for SDN technology adoption are SWAG‘s at best. It’s just too early to know with any degree of certainty how quickly SDN-enabled hardware and software products will find their ways into production deployments and what impact other factors, such as competing technologies and macroeconomic conditions, will have on SDN investments.
Regardless, within three years Gartner will be reporting on the SDN market, if for no other reason that its clients, a large percentage of whom are IT vendors, will demand it. Gartner is, after all, a for-profit business. That is my prediction.