Startup Profile: groundcover Levers eBPF for Monitoring

As cloud-native distributed systems proliferate, observability in the form of application performance monitoring (APM) has become an essential tool for performance and reliability. But a startup named groundcover (which prefers to be known as groundcover without the initial cap) says that adding monitoring to cloud environments can be tough to scale and generate lots of data, with sometimes vague results. And costs grow with the volume of data observed.

“Some companies are spending over a million dollars a year on solutions,” said Shahar Azulay, cofounder and CEO of groundcover, in a recent interview with Futuriom at KubeCon Paris. “We compete by not forcing customers into unexpected fees.”

groundcover’s InCloud service leverages extended Berkeley Packet Filter (eBPF), an open-source software tool that streamlines the speed, efficiency, and visibility of programs that use Kubernetes-based microservices. eBPF does this by running directly inside the Linux kernel instead of on top of it. This gives eBPF the ability to track and even control traffic without burdening the application layer – an essential feature in highly scalable environments. groundcover claims its eBPF implementation via a sensor named Flora outperforms competing observability platforms, including those from Datadog and New Relic as well as ones based on OpenTelemetry.

To keep costs down, groundcover also uses what it calls stream data processing along with open-source database technology, such as ClickHouse and VictoriaMetrics, to selectively store key observability data, reducing the level of extraneous data "noise." This enables all observability data to stay within the customer’s environment – even though groundcover’s solution is delivered as software-as-a-service (SaaS). The result is reduction of “ratcheting fees” associated with some other services that charge according to data volume processed on the provider’s cloud.