Recently I read the article “Why Now Is The Time To Go Serverless” by Romi Stein, the CEO of OpenLegacy, a composable platform company. While I agree with Romi on several points he made around the importance of APIs, micro-service architectures, and cloud computing. I agree that serverless doesn’t truly mean computing without a server, but rather computing on servers owned and provisioned by major cloud providers. My main point of contention is that large businesses executing mission-critical functions in public clouds may eventually come to regret this move to a “Serverless” architecture as it may also be “Trafficless.” Recently we’ve seen a rash of colossal security vulnerabilities from companies like Solarwinds and Microsoft (Outlook Server). Events like these should make us all pause and rethink how we handle security. Threat detection, and the resulting aftermath of a breach, especially in a composable enterprise highly dependent on a public cloud infrastructure, may be impossible because key data doesn’t exist or isn’t available.
In a traditional on-premises environment, it is generally understood that the volume of network traffic within the enterprise is often 10X that of the traffic entering and leaving the enterprise. One of the more essential strategies for detecting a potential breach within an enterprise is to examine; hopefully, in near-real-time, both the internal and external network flows looking for irregular traffic patterns. If you are notified of a breach, an analysis of these traffic patterns is often used to confirm a breach has occurred. To service both of these tasks copies are made of network traffic in flight, its called traffic capture. The data may then be reduced and eventually shipped off to Splunk, or run through a similar tool, hopefully locally. Honestly, I was never a big fan of shipping off copies of a company’s network traffic to a third party for analysis; many of a company’s trade secrets reside in these digital breadcrumbs.
Is a serverless environment also trafficless? Of course not, that’s ridiculous, but are private cloud providers willing to, or even capable of, sharing copies of all the network traffic your serverless architecture generates? If they were, what would you do with all that data? Wait, here’s another opportunity for the public cloud guys. They could sell everyone another service that captures and analyzes all your serverless network traffic to tell you when you’ve been breached! Seriously, this is something worthy of consideration.