KSOC Aims to Create Fingerprints for Container Images

Under an early access program, KSOC has made available a free catalog for accessing open source container image fingerprints that can be used to detect any changes made to an environment.

Based on a tool running in an extended Berkeley Packet Filter (eBPF) subsystem of the Linux kernel, KSOC is making use of a capability, dubbed RAD, to create those fingerprints.

KSOC CTO Jimmy Mesta said while the company is committed to maintaining this repository, it also plans to offer a commercial instance of this capability to enable enterprise IT organizations to apply fingerprints to their own container images.

Mesta noted that the overall goal is to simplify the detection of changes to container images indicative of cybersecurity compromises or, in the future, potentially performance issues.

Only the latest releases of Linux currently support eBPF, so it’s only recently that the level of observability needed to create fingerprints for container images has been viable. However, from a software supply chain security perspective, the implications are profound because any change to a container image in a runtime environment could be used to generate an alert. That approach would thwart cybercriminals attempting to insert malicious containers into cloud-native applications.

In effect, those fingerprints provide a portable approach to application security that can be consistently reused across the entire application environment, noted Mesta.

In general, eBPF is about to transform everything from security and observability to storage and networking. Software that previously ran in user space will increasingly be shifted to a sandbox environment residing in the kernel. That will enable, for example, observability capabilities across the entire application without necessarily requiring agent software to be embedded in every application.

It’s unclear to what degree fingerprints of container images will transform cloud-native application security, but given the inherent challenges of protecting these highly dynamic environments, the need for a different approach has become self-evident. Containers may only run for a few minutes, but at any given time, there might be thousands of them running. Determining which one might have been tampered with is going to be exceedingly difficult to detect in the absence of any evidence. Attempting to detect which alerts are more noise being generated by applications versus an event that is indicative of an actual security breach is currently too difficult, noted Mesta.

In the longer term, it should also be simpler to apply machine learning algorithms to fingerprints of container images to further automate security operations, he added.

It may be a while before fingerprints are pervasively applied to container images, but as in real life, fingerprints provide a unique identifier that might prove to be a critical element of any approach to zero-trust IT. After all, at the root of any zero-trust initiative is any ability to manage not just the identity of end users but also of individual software components.

The challenge now will be determining just how many container images need to be fingerprinted.

Mike Vizard

Mike Vizard is a seasoned IT journalist with over 25 years of experience. He also contributed to IT Business Edge, Channel Insider, Baseline and a variety of other IT titles. Previously, Vizard was the editorial director for Ziff-Davis Enterprise as well as Editor-in-Chief for CRN and InfoWorld.

Mike Vizard has 1641 posts and counting. See all posts by Mike Vizard