Juniper Acquires Apstra, Extends Autonomous Options to Data Center

eWEEK NETWORKING BUSINESS ANALYSIS: Apstra delivers multi-vendor intent-based networking, improving Juniper's overall options.


Juniper Networks this week announced its intent to acquire Apstra, a pioneer in data center intent-based networking (IBN). IBN has been gaining momentum during the past few years as organizations begin to move to fully autonomous networks. 

While most of the mainstream network vendors have some variation of an IBN solution today, Apstra is the only one that works across multiple vendors. Apstra supports most of the mainstream network vendors, including Juniper, Arista and Cisco Systems, but it also supports some alternate vendors such as SONiC, Cumulus, NVIDIA and VMware. This lets Apstra customers enjoy the benefits of IBN without the fear of being locked into a single vendor. 

Although Apstra is a relatively young company, founded in 2014, it does have a number of Fortune 100 customers and service providers, including Bloomberg, Yahoo, Japan, BeElastic and T-Systems, which is a joint customer with Juniper. Apstra was initially designed for large customers, and its solutions scale to thousands of network devices, while its focus on ease of use is a good match to all types of organizations--including mid-size companies--that need network automation. 

Apstra delivers multi-vendor intent-based networking 

Some of the IBN solutions are IBN in name only and designed to automate the deployment (Day 0) and initial configuration (Day 1), but Apstra excels in fully automating Day 2 operations, giving it a product that works across the full lifecycle of network operations. I’ve talked to a couple of Apstra’s customers, and they rave about how easy it is to use the product. Apstra is managed through an easy-to-use GUI that makes deployment simple and does not require expensive post-sales support services. One of the more interesting features is Apstra’s Intent Time Voyager, which enables customers to roll back a portion or all of the network back to previous states. 

Apstra has a rich set of REST APIs that makes integration into self-service, DevOps and other tools easy. Given Apstra’s system is closed-loop in nature, its data set can be used as a single source of truth. Typically, networks--particularly multi-vendor ones--have multiple data silos with multiple management consoles, which leads to the “swivel chair” management approach, in which network engineers have to pivot between multiple management consoles. Apstra applies machine learning to the single data set to predict problems, enabling them to be fixed before they become business impacting.

How Apstra rounds out Juniper’s AI networking capabilities 

For Juniper, this acquisition fits perfectly and rounds out its self-driving network vision. In 2019, Juniper acquired Mist Systems, which had built an advanced AI system to automate WiFi operations. Since then, Juniper took the Mist capabilities and applied them to the wired network, leading to automation of an entire enterprise campus. Earlier this year, Juniper pulled the trigger on 128 Technology, which brought it AI capabilities for the WAN. Juniper already had a strong SD-WAN solution with Contrail, but the addition of 128 Technology has enabled Juniper to shift from SD-WAN to an AI-driven WAN. 

Apstra addresses the data center, enabling Juniper to bring automation across the end-to-end network. This is key, because applications traverse the WAN, data center and campus network. Although these have been managed in silos historically, more and more companies are breaking down these discrete groups to understand how applications perform over the network versus just one portion of it. This is critical in ensuring a continuous, best-in-class end-user and customer experience. Businesses now compete in the experience economy, where poor-performing applications can lead to underproductive users and unhappy customers. 

Apstra plus Contrail leads to automated network fabric automation 

With this acquisition, Juniper can combine Apstra’s IBN capabilities with its Contrail enterprise multi-cloud to automate lifecycle operations of a network overlay fabric. In addition to Apstra and Contrail, fabric automation is made possible because of Juniper’s Junos operating system, which was designed to be open, programmable and provide rich, real-time network telemetry. Although Apstra works with multiple vendors, customers have the option of going end-to-end Juniper with its QFX and PTX switch series. 

It's important to note that Juniper plans to continue to support Apstra’s existing multivendor functionality. The Apstra product is designed to be inherently abstracted from the underlying equipment. That architectural separation will remain. I believe this to be the right strategy for Juniper because it will help them competitively.

Even if the customer is using a competitive vendor, Apstra can give Juniper an entry point, because it enables network professionals to shift to a multi-vendor strategy. Once the company gets a footprint in the account, it can look to extend from Apstra to a broader footprint in the data center, WAN and campus when the network needs to be refreshed. This is much easier than trying to convince the customer to rip and replace the incumbent. 

Juniper has had an interesting history with intent-based networking. The first time I ever heard the term was at a Juniper event perhaps a decade or so ago when the company’s founder, Pradeep Sindhu, predicted IBN was needed in the future to enable networks to continue to scale to cloud demands. Then the company went quiet and fell behind many of the established vendors. 

Since then, the company has been on an acquisition spree that has included Mist, 128 Technology and now Apstra, giving it one of the most complete offerings in the business. I’ve followed Apstra since it launched and expect this acquisition to act as an accelerant in Juniper’s ability to deliver an end-to-end AI-driven network. 

Zeus Kerravala is an eWEEK regular contributor and the founder and principal analyst with ZK Research. He spent 10 years at Yankee Group and prior to that held a number of corporate IT positions.