As Software-Defined Networking (SDN) evolves, its promise is clear: agility. Enterprises and communication service providers alike have been able to significantly accelerate the time to deliver new applications and services as a direct result of software-defined technology.
From an infrastructure monitoring perspective, it also makes correlation of performance events easier. In fact, I recently spoke with Hussein Khazaal from Nuage Networks about this very topic: “With SDN, because the application is network-aware, that correlation [of performance issues] is automatically done,” says Hussein. “If your page refresh is taking too long, you can immediately correlate that to a specific part of the network. And because it’s programmable, you can develop tools to automatically re-route around those problems.”
Despite the benefits of SDN, the technology also introduces new challenges, including its impact on day-to-day performance monitoring.
Challenge #1 – Addressing dynamic real-time change. The ability to automate the provisioning of new converged infrastructures in minutes and impact multiple devices at the same time is a game changer, especially considering that today’s relative static environments rely on manual configurations. With SDN, new compute, network and storage devices and features are immediately available for use. When only running daily checks on what’s new in your environment, these dynamic, real-time changes mean significant gaps in visibility.
What’s needed is a performance monitoring solution designed with open APIs. This way you can integrate directly with SDN systems, listen on the event bus and look for new devices, services or changes, and then immediately modify the infrastructure monitoring inventory to ensure performance visibility.
Challenge #2 – Accommodating rapid on-demand growth.The inevitable uptick in demand for new compute, network and storage in software-defined infrastructure poses a risk to monitoring platforms. These solutions must be able to add monitoring capacity to accommodate the rapid growth of the infrastructure. If they can’t add additional capacity on demand, they can quickly become over-subscribed, creating performance visibility gaps.
Unlike legacy infrastructure in the SDN world we can have multiple overlay topologies running on top of the physical network. Whenever a new service starts, it deploys the necessary virtual infrastructure, and thus the number of monitored elements can grow rapidly with increased demand – outstripping traditional capacity management.
The solution is to deploy performance monitoring within both physical and virtual appliances. When extra performance management capacity is needed, spinning up additional virtual appliances on demand enables performance monitoring to flex with the demands of an SDN environment and still provide answers in seconds.
Challenge #3 – Integrating service context. Having service context is an expectation today. As a result, performance monitoring needs to be able to listen in context of a particular customer or tenant of the network. Ultimately, users should be able to not only ask about the health and performance of individual devices or links on the network, but also, “How is Customer A, HD Video Service: New York to London, performing?”
This also extends to service topology, meaning the controllers and performance monitoring solutions share the knowledge of physical and logical connectivity of the devices – both physical and virtual – that make up a service, both in real-time and for historical context.
Bottom line: SDN is still maturing, and throughout its evolution, it’s important to look at how dynamic real-time change, rapid on-demand growth and integration of service context will play a key role in enabling a successful deployment and avoiding performance visibility gaps in your infrastructure.
For more information, check out this white paper, The Top 3 Challenges of Monitoring Software-Defined Everything.