Juniper CTO talks COVID-19, SD-WAN and shifting multicloud technology

In December, Juniper Networks  named industry veteran Raj Yavatkar chief technology officer. After extended stints with Google Cloud, VMware and Intel, Yavatkar is now charged with setting Juniper’s future courses in intelligent self-driving networks, security, network virtualization, packet-optical integration and how it will support hybrid cloud scenarios. Yavatkar recently talked with Network World about what technologies he sees currently impacting networking and what Juniper is planning for the near future. Here is an edited version of that conversation.

What impact has the COVID-19 pandemic had on Juniper?

First of all we are in a quiet period [ahead of the company’s first quarter earnings results set to be announced April 28] so we’ll talk about that more but suffice to say work has changed.  For example we have made every effort to provide our engineers bandwidth and high-speed internet access to do their jobs remotely. We have seen a high demand for increased security as well as bandwidth for employees working from home, but in general people are coping with the situation. 

What has been the focus of some of your recent technology directions?

We have been focused on a couple things, one making sure we get traction on our secure SD-WAN software components [Juniper rolled out its Contrail SD-WAN cloud package in December to include support for SD-LAN-specific operations, such as provisioning of new devices and managing branch office LANs].   And we are looking at providing more multicloud connectivity and building on our ability to provide cloud-agnostic connections to data centers, SD-WANs and branch offices. There are a ton of opportunities there.

What’s the status of Mist integration with Juniper’s portfolio?

Mist [Juniper in April closed the agreement to buy wireless-gear-maker Mist for $405 million and has been incorporating the Mist technology with its own] has exceeded our expectations, and we are seeing some large companies go with the Juniper/Mist technology in particular to extend Mist’s AI techniques for self-diagnostics to data-center fabrics.  Mist’s AI engine, Marvis, [which is known for its cloud-managed, AI-based wireless service called WiFi Assurance, which measures performance and service-level metrics to make wireless networks more predictable and reliable, which also features  machine-learning technology to automatically identify, adapt and fix network issues] does a great job in terms of handling wireless access point monitoring and cell remediation, but we are extending it beyond that to help reduce trouble tickets in data-center operations for example.  The AI and automation can reduce trouble tickets by half, which significantly reduces the load on IT.  

Juniper was among the first vendors to roll out 400G Ethernet products. How is that going?

We have started shipping it and have a number of ongoing customer trials.  Right now we are seeing hyperscaler, cloud providers and larger enterprise interest mostly so far. We expect that will grow, but COVID-19 might impact things – that remains to be seen of course.

Looking forward, what trends you see impacting enterprise networks?

There are a couple things we are looking at. If you look at the work Google Cloud is doing, getting more of the application layer involved with microservices through a service-mesh proxy is changing the way we can develop applications and the network.  The idea is to improve the interaction between the overlay application layer and the underlying network layer. 

The other big area we are working on is automation. Enterprises need a lot more automation. When I was with VMware and Google, we developed automation for everything, and that’s where the data center is going especially in a multicloud world.  We can gather advanced telemetry and analytics, and feed into advanced automation capabilities to create a powerful data-center fabric.  It’s a logical next step for Juniper.