So what’s the holdup?

Three main areas of interest aren’t being talked about at the moment, but all of them represent clear obstacles to the successful deployment of SDN:

• Islands of SDN. Each vendor will implement the standard in a slightly different way or implement a different version of the same standard. And since no real interoperability events have been held yet, it appears likely that each vendor will represent an “island of SDN” in its own right.
• Islands of differing topologies. No deployment will be completely based on SDN alone but rather on multiple topologies, including legacy, hybrid, cloud, and virtual networks.
• Multitenancy. Competing companies using the same SDN infrastructure should know that the network itself is being monitored to ensure the separation of “church and state,” thus delivering the level of trust and proof the end user needs to deploy an actual SDN solution in a multitenancy environment.

The obvious challenge here is monitoring different network types that aren’t necessarily interoperable and that reside on multitenant servers. By introducing a seamless monitoring capability that encompasses all the leading SDN infrastructure vendors, Gigamon hopes to help the entire industry move forward.

The path to follow

When rolling out SDN, a highly agnostic monitoring strategy is needed to better help vendors and their end users get it right the first time and experience a smooth deployment cycle.

SDN deployments require high-level monitoring capabilities that allow for a greater reduction of monitored traffic through advanced and multithreaded filtering as well as packet manipulation. This, in turn, opens the door to greater integration with analytic tools while enabling those tools to perform more effectively by maximizing their analytic throughput.

Simple tap functionality is now just a starting point. A monitoring network that makes SDN deployments possible, not to mention traditional and hybrid deployments, will need to supply much more functionality at the packet, flow and network-wide levels to truly bring about the visibility that’s required. Visibility services such as multidomain flow mapping, packet normalization, flow optimization, and de-duplication will form the basis of the monitoring fabric.

Because SDN will be deployed in an early, multivendor fashion across different network topologies, finding a way to gain overarching visibility into the traffic on CSP networks is essential. To date, no such visibility and monitoring capability exists, but given the perceived issues with interoperability, this problem will need to be solved sooner rather than later in the life cycle of SDN.

One possible solution would be an applet that ties together flow-routing data from various vendors’ control points and from legacy, hybrid, cloud, and virtual networks. Yet again there has been minimal discussion so far about this desired capability, but the need for it is very real.

From a CSP’s perspective, SDN is the path to follow for several reasons, from commodity hardware to new ways of controlling services to sheer acceleration of data across a network. The obstacles could ultimately dissuade a provider from realizing these advantages, but Gigamon sees great value in SDN technology, and with the provision of a monitoring infrastructure, the speed of adoption could be rapidly increased.

A number of network equipment vendors have recently introduced SDN infrastructure components, and deployments have already begun to happen at the tier 1 level for CSPs, primarily to add flexibility to existing service offerings. But as in previous tech cycles, “the next big thing” is in danger of running aground if the right monitoring capability doesn’t accompany the rollout. Only time will tell.