Cumulus Targets Telemetry & Microservices


By: Michael Vizard

Cumulus Networks has released version 2.0 of NetQ of its network operating system (NOS), focusing on programmability and networking visibility features.

NetQ now supports Kubernetes and a set of REST application programming interfaces (APIs) to boost management capabilities. In addition, Cumulus has added support for gathering data in real time, richer analytics, and diagnostic capabilities.

Finally, Cumulus Networks has revamped the user interface of NetQ to make it easier for network administrators to navigate NetQ.

Increased Visibility

Cumulus Networks is trying to make it simpler to identify potential issues relating to the NOS environment spanning everything from the container, virtual machine, and host platforms, says Partho Mishra, vice president of engineering for Cumulus Networks.

Such features have become more critical, thanks to the dramatic increase in DevOps applications being deployed and updated. Network administrators are simultaneously trying to cope with many issues, including microservices-based applications running on platforms such as Kubernetes. These approaches make the overall networking environment more complex to manage because each distributed instance of a microservice needs access to network services via an API.

“There’s a lot of intense pressure,” says Mishra.

In addition, Cumulus is looking to build a NOS itself that is made of discrete microservices that will provide more flexibility in terms of where those network services will need to be deployed. As a first step towards achieving that goal, Cumulus is exposing REST APIs that will make it simpler to remotely manage those networking services. That architecture will also make it simpler for Cumulus to continue adding new functionality without having to upgrade an entire monolithic operating system.

Network-DevOps Convergence?

As more APIs get exposed for software purposes, it's becoming easier for developers using a DevOps approach to build in networking services to applications. But it’s still not clear to what degree network operations will meld with DevOps. Some proponents argue that IT organizations will only need to expose network services to the degree that developers can self-service their own needs within guidelines defined by network administrators. Others argue that networking is so core to application performance that developers will want to be able to dynamically control networking services on a more granular level.

Depending on the nature of the application, both requirements are likely to arise. The challenge IT organizations will face is determining what level of integration between NetOps and DevOps makes the most sense based on the attributes of the application workload.

In the meantime, there’s a clear race on between networking vendors to deliver a NOS based on modern microservices-based architectures. It’s going to take a while before most organizations have the processes in place to fully appreciate the capabilities of those NOS platforms. But the one thing that is clear is that the day of the monolithic operating systems, tightly coupled to a narrowly defined set of switches and routers, is finally coming to an end.