SDN in the Enterprise: aligning with business needs
There is always a well-known solution to every human problem–neat, plausible, and wrong. – H.L. Mencken
As a long-time practitioner of the art of beating computers and communications systems into submission, I am as enamored with the latest gee-wiz technology trends and tools as the next self-respecting geek. I’m also not completely above the allure of the herd-mentality; all for one and all for the new tech. As an IT Director looking at the business side of the house, however, and having to translate all of the latest trends into actionable business intelligence and strategy, I am far less quick to jump on the latest bandwagon. Sometimes what my cohort are talking about, and what I find fascinating personally, isn’t what the business needs. Often, it’s not even close.
It can be a challenging thing, trying to match potential technology solutions to existing or future business problems. It can be even more challenging separating the latest trends and market buzz-word bingo, from the actual solutions that will help my company move forward. Finding those solutions can sometimes seem like a search through the proverbial haystack.
Other times, a solution is fairly self-evident and matches existing capabilities that can be implemented today. The market is talking about these things, vendors are selling these things, and people like me are solving problems using these things. The Nexus product line from Cisco would be a good example of this. It came out with features that we wanted, that solved specific problems we were having at the time (10-Gig port density), was available now, and so after some investigation we began the process of planning for implementation.
Sometimes, however, the blogosphere lights up with palpable excitement at some new thing that isn’t a product, or available, or even well-defined. In some cases, the new “thing” is so exciting and so revolutionary that all reasonable sense is lost and even the most jaded of tech analysts start predicting massive market shifts, even assigning dollar values to future years for the emerging technology.
While “cloud” and “DevOps” are recent offenders in this category, I’ll just pick on the latest “thing-du-jour” to come out of the ether: Software Defined Networking (SDN).
SDN is, in many ways, the natural follow-on to the abstractions that have taken place in the 10 years or so in the compute space with virtualization.
As a quick aside, by the way, while most of us think of virtualization and hypervisors as a relatively new concept–when I started converting our corporate data centers in 2006 a lot of my peers thought I was crazy at worst and buying into a fad at best–it has been around a lot longer than most people realize. Check out hypervisor on wikipedia for more information, and be sure to follow the references.
The general idea in virtualization of compute resources was that bare-metal servers were inefficient since at any given time there were apt to be large amounts of wasted resources. Virtualization helped (in some ways) with that by separating (abstracting) the underlying storage, processor and memory pieces from one another and from the operating system using them. This market continued to mature and grow, and the industry as a whole came to accept this, largely, as the new paradigm for compute (at least on the server side with predictable and amenable workloads; corner cases, desktops, etc. is another discussion).
During all of this growth and acceptance, however, the network piece largely stayed the same as it always had been, with few exceptions (Cisco with the Nexus 1000 for instance). Enter SDN which, ostensibly, aims to abstract the underlying network pieces (forwarding plane, control plane, etc.) from one another and allow the same kind of flexibility on the network side as we now can have on the compute side. The problems, however, are not the same and not being grown in the same manner as in previous cycles. Therein lies many of the problems.
In the initial cycles, you really had one company (VMware) driving the virtualization bus. They brought the technology to the fore, penetrated into the market with it, and brought a tangible, quantifiable, explainable benefit to a significant number of businesses. The market was, and largely still is, defined by VMware in the context of their technology.
In the SDN market, you don’t have one entity pushing a single vision of the future, you have many, many, many. You have individual manufacturers hurriedly bolting on something (look, we have OpenFlow-enabled switches) to their existing product-line, even in the absence of market-wide acceptance of their definition (shades of “Next Gen Fiber Channel” here); you have different, competing groups trying to define the whole market or pieces (OpenFlow, OpenStack, Open-can-of-worms); and you have VARs and thought leaders going completely, indiscriminately crazy with excitement at this thing which mostly exists as an idea, with bits of product around the edges.
Another aside: check out J Metz’s excellent write-up on the Marketecture of Next-Gen Fiber Channel
In the business of IT, we often forget that we are business enablers first and foremost. We exist because the business has needs that are fulfilled by the IT organization, not because the business necessarily wants this wonderful, cutting-edge new technology. Everything we do is measured against, and justified in the context of, business results.
In the next piece of this segment, I’ll explore what I think the business drivers of SDN might be, what they might not be, and how any of this might change the IT organization. I’ll also talk about how I see careers changing (or not) as a result of this new technology, and why I don’t see this as being quite as disruptive or earth-shattering as others.
In the mean time, I’d love to hear your feedback. Tell me why you think I’m right, or wrong in my analysis, and what you think about SDN as a whole.