Secure access service edge (SASE) saw rapid growth during 2020 as enterprises grappled with the abrupt and unceremonious shift to remote work in the wake of a global pandemic. But while the technology addressed many of the security challenges associated with a remote work force, it didn’t necessarily solve the problems faced by the workers themselves.

Dropped Zoom calls, applications running at a glacial pace, and bad WiFi have become a fact of life. And these challenges have put a new burden on IT teams tasked with keeping employees connected and productive.

“All the user knows is that ‘my Office 365 access is not working,’ and I have an IT staff that’s generally a small team trained to troubleshoot corporate networks in an enterprise office environment. Now they have all these home users and a huge number of them are calling in asking for help,” explained Craig Connors, CTO of VMware’s SD-WAN and SASE business, in an interview with SDxCentral.

It’s at this juncture that companies like VMware hope artificial intelligence operations (AIOps) platforms can help.

VMware’s SASE Gets Smart
In retrospect, VMware’s timing couldn’t have been better. The company announced the acquisition of AIOps vendor Nyansa a year ago this week. Since then, Nyansa’s Voyance platform, now rebranded as Edge Network Intelligence, has become an integral part of VMware’s SASE.

“The acquisition of Nyansa was made in large part to bring AIOps and this intelligence into SD-WAN to facilitate a move from point-to-point error correction methodologies that we have applied historically in SDN to a broader end-to-end, network-based approach,” Connors said.

Prior to its integration with VMware, Nyansa could pull in real-time and historical data from devices on the network. This information was processed through a series of machine learning and AI algorithms designed specifically to identify and remediate network disruptions.

“Using all of that data, they could build a very robust set of information about what problems were occurring and where on the LAN,” said Connors.

VMware has since taken these capabilities and integrated it into its SD-WAN appliance, he explained. “And because it’s a border gateway between the LAN and the WAN, we’re able to not only detect when application problems and connectivity problems are happening, but we’re able to isolate the fault to whether it is a LAN problem or WAN problem.”

This, he said, is critical because most of the issues remote workers are facing today aren’t actually with their WiFi or LAN. “It’s the fact that when a user has a problem and they call the help desk, the help desk is really at a loss for how to help them troubleshoot the problem and fix it.”

To further empower these capabilities, VMware has begun integrating with key applications like Zoom to gather metrics on call performance. This allows users to get a better picture of how applications are performing.

These capabilities feed into an anomaly detection engine that can help to determine whether there is a LAN issue affecting a single user, a hardware issue affecting users with a specific piece of equipment, or a WAN issue affecting an entire region.

And those insights aren’t siloed to one enterprise customer, said Connors.

“We have this huge user base that are all going through our gateways, so you may not be on a Zoom call yet when we identify that users in your area traversing a certain gateway are having Zoom issues,” he said, adding that traffic can be rerouted around the disruption before an issue occurs.

AIOps Is Becoming a Differentiator
VMware isn’t the only SD-WAN, or even SASE vendor, employing machine learning to identify and remediate disruptions. Open Systems, another SASE vendor, made several acquisitions early last year to bolster its AIOps capabilities, particularly around security.

Meanwhile, both Aruba and Juniper — two companies that recently entered the SD-WAN market with the acquisition of Silver Peak and 128 Technologies respectively — have made steady investments in AIOps capabilities to combat network disruptions and resolve application performance challenges.

However, Connors argues that VMware’s ability to draw in data from its points of presence and use that information to reroute traffic around disruptions is what sets it apart from other vendors.

“If you’re not able to reroute folks through the cloud or apply error corrections across the last mile and give someone a high-quality handoff into Zoom, you can’t fix Zoom problems,” he said, adding that VMWare’s distributed SD-WAN gateways and SASE architecture allows the vendor to do just that.

Microsoft launched a more complete Kubernetes tool to manage clusters on its Azure infrastructure-as-a-service (IaaS) that its open source team recommends as a replacement for the existing AKS Engine offering. The Cluster API Provider for Azure (CAPZ) also takes a step back from what has been a more abstracted view for managing container clusters on public cloud infrastructure.

Craig Peters, principal program manager for Azure container compute at Microsoft, explained in an email to SDxCentral that CAPZ is a tool that can be used specifically on and is complementary to infrastructure like Azure Kubernetes Service (AKS), Azure, and Azure Arc.

It’s built on the open source Cluster API (CAPI) Kubernetes sub-project, which is designed to provide declarative APIs and tooling to simplify provisioning, upgrading, and operating of multiple Kubernetes clusters. CAPZ takes that base and works through the Azure Resource Manager (ARM) to provide greater control for self-managed Kubernetes clusters on Azure and Azure-managed Kubernetes clusters.

“CAPZ enables users to use CAPI to create and manage either self-managed clusters where the user owns the control plane, uptime, and maintenance entirely themselves,” Peters wrote. He added that CAPZ is a tool that can be used specifically on Azure infrastructure through ARM to manage Kubernetes clusters through the Cluster API.

Basically, CAPZ provides for a greater level of management over Kubernetes clusters that are running in one of these managed environments. This allows it to bridge the gap between a user relying exclusively on their managed host for that control or attempting to manage Kubernetes on their own.

In an accompanying blog post, Peters explained that while AKS remains a “mature, scalable, secure” and fully Microsoft-based platform, it does not have the depth of control that some developers require.

“Some need functionality that is not available in AKS yet or might never be because they require user access to the control plane,” he wrote, noting that this need could be tied to regulatory or compliance issues, or even just a developer needing more control over Kubernetes features.

CAPZ Caps AKS Engine
That greater flexibility is also behind Microsoft’s move toward recommending CAPZ over its legacy AKS Engine product. AKS Engine is designed as a templating tool for turning cluster models into ARM templates to manage Kubernetes clusters but lacks flexibility as it can only create one cluster at a time, which limits scalability, and only on Azure.

“AKS Engine has been the workhorse for teams who need to operate their own clusters on Azure,” Peters explained. “The ARM template workflow is natural for some Azure-focused teams but doesn’t provide the power and flexibility of CAPI.”

However, Peters’ in the blog post noted that the “design falls short of empowering ongoing operational needs such as scaling, in-place upgrading, and extension management. And it isn’t useful for users who are focused on multi-cloud scenarios like managing fleets of Kubernetes clusters across cloud infrastructures that do not support ARM.”

As such, his team is suggesting that AKS Engine users look at CAPZ as it provides stronger support for managing the cluster lifecycle and that it will be the focus for all new investments from that team.