Four elements of high-availability and resilient cloud networks

COMMENTARY: Nearly every organization now consumes some form of cloud computing in its routine operations. Today, cloud services have become the lifeline of digital business continuity.

But what if the lights go out? It's exactly what the world's businesses had to confront when major hyperscaler AWS experienced an outage on October 20.

[SC Media Perspectives columns are written by a trusted community of SC Media cybersecurity subject matter experts. Read more Perspectives here.]

Services went dark in cascading waves, from messaging services and payment systems to fast food chains, forcing the tech community to confront an inconvenient truth: hyperscalers are fallible.

And, it's not just AWS.

When an outage struck the Google Compute Platform (GCP) in June, infrastructure services that rely on GCP went down for hours.  High availability isn’t optional: it’s essential today. And achieving it starts with architectural ownership, leveraging infrastructure an organization fully controls.

Mission-critical functions simply can't afford to be offline. But when they're built on hyperscalers such as AWS, Azure, or Google Cloud, they inherit not only the scale and convenience, but also the fragility of those systems. A glitch here will compromise access, render systems vulnerable, and erode trust.

Critical service providers need to own and control their own global, private infrastructure and Points-of-Presence (PoPs) to establish an unfailing backbone and redundancy that doesn't fail when hyperscalers let them down.

The need for high availability

When critical infrastructure such as SASE gets hosted on hyperscaler networks without ownership of the underlying infrastructure, an outage can cut off access to apps and disrupt identity and access management workflows. This leaves many organizations with no fallback paths. Secure web gateways (SWGs) and firewalls may stop enforcing policies, and incident response tools may become inaccessible, dangerously ruining the customers’ security posture.

Unlike providers that own their infrastructure, hyperscaler-dependent services lose visibility into the underlying network and lack control to enforce their resilience strategies. For example, they are unable to instantly shift workloads across providers or maintain session persistence during failover. With employees and customers expecting seamless access, user trust is on the line.

High availability has become an underlying requirement for modern enterprise functions that must simply work all the time. It makes sure that main services are up, no matter what component- or provider-level outages may occur.

The advantage of owning the core

A global, private backbone of PoPs can introduce a whole new level of resiliency. It doesn't just sit idly by waiting for the cloud to come back online. It keeps on redirecting traffic, enforcing policies, and providing users with an unbroken, seamless experience.

Having infrastructure ownership allows for instant failover between regions or providers, making it less dependent on any third-party cloud provider. Having extensive visibility into traffic and latency measurements enables informed rerouting decisions. Visibility and control over every tier of the tech stack offer consistent security and performance.

Build resiliency

Here are the four elements of truly high-available and resilient networks:

Architecture with redundant layers: Engineer networks with redundancy at each layer by having multiple PoPs within each region, and dual or triple links between key hubs. This way, when one component fails, another takes its place immediately, without any perceivable decrease in performance.

Dynamic traffic redirection: Dynamically reroute traffic through alternate PoPs, backbones, or between cloud providers. This typically occurs automatically and in milliseconds, maintaining user experience even in the middle of an outage.

Decentralized policy enforcement: Implement security and access policies at the PoP level instead of redirecting all traffic to a centralized engine. This minimizes backhaul, eliminates the need to reconfigure policies, enhances performance, and provides protection even if a centralized node fails.

Automation and self-healing infrastructure: Adopt automation for resiliency. Apply self-healing methods that detect node degradation and re-route workloads. Automated patching and upgrading prevent downtime due to human error.

What enterprises should ask their vendors

Companies need to collaborate with services that combine hyperscaler scale with independently owned infrastructure, allowing them to provide uptime when giants fall. Here are a few questions that enterprises must ask before choosing a vendor:

Outages aren't a question of if, but when. Critical infrastructure constructed solely on hyperscalers only remains available as long as the hyperscalers function. Services that have network ownership, failover automation, and distributed security enforcement can create real-world resilience that endures beyond any hyperscaler hiccup.

Aviram Katzenstein, chief platform officer, Cato Networks

SC Media Perspectives columns are written by a trusted community of SC Media cybersecurity subject matter experts. Each contribution has a goal of bringing a unique voice to important cybersecurity topics. Content strives to be of the highest quality, objective and non-commercial.