SDN and the Network Visibility Challenge
Despite the hype, the enterprise remains reluctant to adopt software defined networking. The experts may claim SDN will be amazing, but the experts aren't responsible for the day-to-day health and safety of the data environment that current enterprise infrastructure supports, however imperfectly. CIOs and top IT managers are, and they have a hard time trusting what they cannot see. In a complex and constantly shifting software defined network that might comprise all manner of physical, virtual, on-premises, and/or hosted infrastructure, what they cannot see could add up to a lot.
At the moment, unfortunately, no effective ways exist to measure crucial SDN aspects like end-to-end user response times, or to isolate and repair issues on the public WAN. As JDS Uniphase’s Rob Marson puts it, “How do you detect and resolve issues in the network if it’s not actually your network?” SDN management regimes, therefore, need to look deeper than packet headers and the TCP/IP stack. They need to peer directly into application content. They also need to be flexible and scalable enough to remain in tune with the dynamic nature of SDN.
According to IT management firm SolarWinds, workload mobility will complicate management of not only network infrastructure, but storage, I/O, and the applications themselves. Most management stacks are still geared toward silo-based architectures. This makes the need for automated, cross-platform systems a top priority as SDN architectures unfold. The good news here is that even before the vast majority of enterprises take their first steps toward SDN, they should already have acknowledged the need for broad network visibility. Many vendors are developing solutions to address that need.
Cloud-based SDN providers are teaming up with network management firms to provide tools for real-time visibility. Arista Networks, for example, recently paired up with ExtraHop to devise the Persistent Monitoring Architecture, essentially a mash-up of Arista’s Data ANalyZer (DANZ) module and ExtraHop’sContent and Correlation Engine. The system provides agentless visibility across L2-L7 network, infrastructure, application, and transaction environments, as well as server and address auto-discovery, real-time session and flow assembly, and vMotion data migration management. Providers like Arista know that to gain market share, they need to make integrating their services with legacy enterprise environments as easy as possible.
A number of start-ups are keying in on the SDN management space as well. A company called StackDriver, for instance, uses Big Data techniques to analyze data from internal and external infrastructure to provide a cohesive view of the overarching data environment. The company has devised its own analytics engine to detect potential trouble spots, with the added ability to track changes to network configurations as virtual and cloud-based systems are deployed and decommissioned.
Still, is any of this enough for IT executives to get past the trust issues regarding SDN and the cloud? Perhaps not right away. But as the comfort level with outside resources grows, the management and visibility tools that provide truly effective service will rise to the top. Broadly diverse infrastructure can then quickly become the norm rather than the exception.
And once we’ve reached the point where mission-critical workloads find their way onto the cloud, it will be good to know that there are tools available to foster the same operating philosophy that governed the waning days of the Cold War: Trust, but verify.