Optimizing SDN for the Coming Data Deluge
SDN is the saw, Big Data the wood. What needs to happen before a grand new structure can be built?
There are two ways of looking at the state of enterprise technology today. Either challenges like Big Data and the Internet of Things are fueling the development of the cloud and software defined infrastructure, or cloud and SDx development are driving Big Data and the IoT. Either way, these twin forces affecting Information Technology have at least given the enterprise a goal and the means to attain it.
But is it really that simple? Does simply layering an abstract architecture atop a physical resource enable IT to handle the challenges ahead? Not exactly. As executives at even the most forward-leaning enterprises will tell you, much work must be done to enable coordination and orchestration after the basic SDN environment has been laid out.
Big Data and the enterprise network stack
For one thing, Big Data and other emerging data initiatives require full support of the entire network stack – from physical switching, processing and storage to top-end applications – and this has to be carried throughout an increasingly distributed data environment. This will require a lot of rhythm between discrete vendors and platforms, a fact not lost on many of the smaller firms who hope to leverage Big Data into stronger footholds in the enterprise and cloud provider markets.
The recent alliance between networking firm Plexxi and scale-out server manufacturer PSSC Labs is a case in point. The two have joined forces to create an end-to-end Big Data solution capable of deploying heavy workloads across racks and even widely distributed data centers. The platform combines Plexxi’s Big Data Fabric with PSSC’s CloudOOP 12000 server capable of running Cloudera-certified Hadoop workloads. By tying a network fabric directly to the server, the companies say they can remove many of the bandwidth and connectivity issues that inhibit high-scale operations.
SDN and the Internet of Things
SDN will prove even more valuable when the enterprise confronts the IoT, say Enterprise Networking Planet’s own Julie Knudson. With countless devices streaming data back to the data center on a continual basis, today’s rigid architectures do not provide the flexibility to direct it to the appropriate resources in an efficient manner. SDN not only provides the dynamic configuration capabilities to handle a diversity of data and analytics requirements, but also enables local resources to link up more easily to the cloud, where much of the offloading is likely to occur once Big Data processing gets going in earnest.
Interoperability is just the beginning
Interoperability is also a key facet of this emerging data ecosystem, which is why many organizations are turning to open platforms like OpenFlow and OpenStack. Of course, there are different degrees of “openness,” and many proprietary platforms are gaining toeholds in community-driven open environments with the intent of providing more integrated, streamlined architectures. As BTI Systems’ Mike Marshall notes, a fully interoperable networking stack provides a great deal of flexibility, but as long as open APIs and standard protocols are in place, the network should support the automation and programmability functions that enable interoperability across multi-vendor deployments.
But even this will not be enough to truly incorporate Big Data into the enterprise environment, says Pica8’s Manish Singh. Technology is fine for moving, processing and storing data, but to derive real value you’ll need to put Big Data in context. On the networking front, this involves correlating Big Data events and protocols back to network events and protocols so that workflows can be administered properly and potential trouble spots can be identified and corrected. This will require close coordination between monitoring, visibility, network programming, and a host of other disciplines, but it should ultimately convert today’s static, reactive topologies to more dynamic, proactive ones.
Enterprise networking is undergoing the same changes that are affecting server and storage infrastructure, and with the advent of SDN, it can finally employ the same virtual abstraction and all the flexibility benefits that go with it.
But even the sharpest saw does not a carpenter make. And the challenge, as always, is for the enterprise to leverage the tools at hand to produce the most desirable outcome. If SDN is the saw and Big Data is the wood, it will be interesting to see what kind of structure emerges once the real work gets underway.
Photo courtesy of Shutterstock.