Jul 30, 2015
A lot has been written over the last year about how SDN will fundamentally change how enterprise WANs are architected. While I generally agree with that statement, I think it is important to not get too far ahead of ourselves. Before we implement any new networking technology or architecture we always have to answer a few basic questions such as: What are the key characteristics of the current networking environment? What factors are driving organizations to consider making a change to that environment?
I recently published a research report entitled the 2015 State of the WAN Report that was based on a survey that I gave to over 100 IT professionals. One of the topics that the WAN report examined was the amount of Internet traffic that was backhauled. The survey results show that there is a bipolar approach to how Internet traffic is handled. There are some organizations that do absolutely no backhaul. These are primarily SMBs that have relatively few sites and use the Internet as their WAN. Additionally, a quarter of the respondents indicated that they do limited backhauling. Altogether, over one third of the organizations surveyed in total do either limited backhaul or none at all. At the other end of the spectrum, fully half of the survey respondents indicated that they do significant backhauling, with 40% indicating that they backhaul more than 80% of their traffic.
In the current environment, the typical WAN is comprised primarily of two services: the Internet and MPLS. With that in mind, I asked the survey respondents to indicate which type of applications were driving the biggest increase in their use of those services. It was not surprising that the biggest driver of an increase in the use of the Internet was to provide access to public cloud applications and services. What initially surprised me was that providing access to public cloud applications and services was also a major driver of increased MPLS traffic. That surprised me because few organizations connect directly to a public cloud provider via MPLS. It took me a while to realize that the reason access to public cloud providers was driving a major increase in MPLS traffic is that as described above, half of all network organizations backhaul a significant amount of their Internet traffic on their MPLS network.
One of the key topics that the WAN report explored was the factors that are driving change in the WAN. According to the survey respondents the top 5 factors, in decreasing order of importance, are:
One thing that fascinated me about the survey results was that reducing cost came in fairly low in terms of the factors driving change. That is not usually the case and I attribute its relatively low importance to a couple of factors. One factor is that the economy is generally healthy, and as a result some of the economic pressure that network organizations have felt for a number of years has been reduced. But I don’t think the healthy economy entirely explains why cost is where it is on the list. I think that cost is where it is in large part because there is a broad understanding that the issues listed higher than cost in the list can’t be ignored.
I have yet to run into a network manager who looks me in the eye and says something like “I have always wanted a software defined WAN (SD-WAN).” The vast majority of network organizations don’t have the interest — or the time — to explore new technologies or architecture just because they are the latest thing. In order for SD-WANs to be successful in the marketplace they have to address the factors that are driving change in the WAN and they have to address those factors in a way that is notably better than the current approach.