One of the primary reasons that VMware acquired SD-WAN vendor VeloCloud last November is because it was able to converge the management of policies across an extended enterprise, according to Tom Gillis, senior vice president and general manager of networking and security at VMware.
NV software such as VMware NSX enables IT organizations to centralize the management of policies all the way out to branch office, Gillis said.
That approach, said Gillis, makes it possible to separate the management of policies across a heterogenous networking environment from the packet transfer level. Otherwise, applications wind up having to navigate a series of network tolls to be checked against multiple policy engines.
“People want flat, fast efficient networks,” Gillis said. “That means the best place to enforce policies is at the application virtualization level.”
Flatter networks are critical to make networking teams more agile at a time when DevOps teams want to be able to access network services on demand. Networking teams can now centrally provision those resources without having to reconfigure policies that would otherwise need to be manually implemented one box at a time using command line interfaces (CLI).
A recent survey of 400 IT DevOps and NetOps professionals conducted by F5 Networks and Red Hat finds more IT organizations realize manual provisioning of networks is unsustainable. The survey finds 46 percent of network operations (NetOps) teams have to some degree already adopted DevOps process to become more agile. In fact, the survey shows only 30 percent of NetOps teams are employing multi-vendors tools to automate network management. The rest continue to rely on single vendor tools, including 43 percent that still rely on a CLI or device script.
While NetOps teams have not embraced automation to the same level as the rest of their IT colleagues, they may not be getting enough credit for how far along they really are, said Cindy Borovick, director of business intelligence for F5 Networks.
The survey finds that equal percentages of DevOps teams (43 percent) say deployment frequencies are “good enough” and “not frequent enough.” NetOps were equally split, with 31 percent saying deployment frequencies were “good enough” and 30 percent saying that frequencies did not meet expectations. In a previous study a year ago only 18 percent of NetOps teams indicated that deployment frequencies needed to improve.
“Things are a lot better than they were a year ago,” said Borovick.
NetOps teams find it challenging to provide DevOps teams with self-service capabilities when it comes time to deploy their applications, said Chris Wade, CTO of Itential, a provider of low-code tools for programming heterogenous networks. Manual approaches to managing networks clearly don’t scale. But most NetOps teams are not going to develop extensive programming skills. They need approaches to automation that enable them to work more closely with DevOps practitioners within the context of a larger automation initiative, Wade said.
“We’re starting to see more automation teams developing,” he added.
One thing that is clear. Networking at distributed scale is becoming more automated. The challenge facing IT organizations now is developing new processes that better reflect the current state of the management tooling art.