Believe no single vendor that claims to provide you SDN. Software-defined networking is not a result. Rather, it is a set of principles by which one architects a network to create a more intelligently connected infrastructure that provides increased agility. Put another way, a more agile network is the result of an SDN architectural approach.
There’s been much campaigning about control plane separation and programmability being the critical, key elements for delivering on the SDN vision. Granted, SDN without some form of programmability is impossible. However, it’s the implementation and integration of said programmability that is what’s really important. An API on its own is as useful as a car with no driver—it’s potential without execution.
Acknowledging that SDN is a practice and not a result, let’s now consider the many roles and responsibilities of a typical data center network. First, we have connectivity, both physical (ports, backplanes, throughput) and virtual (distributed fabrics, VLANs, encapsulation). These connectivity-focused elements of the data center network are concerned only with ensuring that a packet can reach the next element in a chain. Improving the agility of connectivity alone doesn’t create a production-ready environment, and much of the human latency of deployment is still present.
So, the question I’m forced to ask is, why limit an SDN project to deliver only part of the solution? A leading driver of enterprises’ interest in SDN is reducing time to value—that is, the time it takes for an asset to be deployed and start generating value to the business. Whether it’s packaged software like Microsoft SharePoint, for example, or a bespoke solution developed internally, delays in deployment of these assets are frustrating for line-of-business owners and, in many cases, a costly problem. Having an asset that’s known to improve productivity or increase revenue sit in a queue makes little sense to anyone.
What does this have to do with partnerships? Well, consider what would happen if you applied SDN principles individually for each type of service—security, availability, etc. This would require separate controllers from various vendors for each service (access federation, application firewalls, compliance, etc.), which would result in administrative issues and, most dangerously, new administrative silos. Remember, the driver for increasing network agility is reducing time to value, which includes solving the human latency problem. Management silos do not solve this problem; they simply move the problem elsewhere. Take a look at this piece in Forbes showing us that the problems associated with management silos are not new; they have been witnessed in business well before becoming a technology problem.
We can only reduce human latency through validated vendor partnerships, integration of management tools, and the reduction of (or, ideally, eradication of) management silos. Succeeding with your SDN implementation requires one to seek out vendors that have partnered—not just on paper, but in combined vision—to decrease administrative time at the keyboard. An additional benefit from such partnerships is the reduction of risks associated with managing many point solutions. Most outages these days are still caused at the keyboard, with IT staff pushed to work harder and longer hours to improve business and reduce time to value.
If you are looking to benefit from the application of SDN principles to your next architectural design, focus on achieving:
- Faster time to value for new applications and services
- Faster time to react to both planned and unplanned circumstances
- Reduction of risk through the elimination of management silo’s
Be wary of any single-vendor SDN pitch that claims to achieve this all on its own. While it may provide some elements of the aforementioned principles, the greater benefit is achieved through an integrated solution.
SDxCentral DemoFriday ALERT: Facilitate & Automate L4-L7 Services in Cisco ACI Environments Using F5 BIG-IP & BIG-IQ. Join SDxCentral and F5 & Cisco on Friday, March 13th to see demonstrations on how Cisco ACI environments can be optimized using the F5 BIG-IP and BIG-IQ management platforms. Register here!