This time I sat down with Karim Traiaia, the Co-founder of Kerno, a company that helps troubleshoot cloud applications. Here are some of the topics we discussed:
The spark
Every startup starts with a pain point—how did Karim come up with Kerno, and what problem was he trying to solve at the beginning?First customers
Landing those early paying users is always tough. Did Kerno rely on free trials, community outreach, or partnerships to get started?The eBPF foundation
Was eBPF part of the plan from day one, or did the idea for Kerno evolve into it over time? And why eBPF specifically over other tooling?What only eBPF can do
Where has eBPF been able to collect critical data that simply wasn’t available from user space or other observability tools?Taming the telemetry flood
Observability tools generate a LOT of events—from syscalls to protocol traces. How does Kerno decide what data to keep vs. what’s just noise, and how much is actually useful in practice?The observer effect
What about CPU and memory overhead—how does Kerno make sure the eBPF agent itself doesn’t impact the workloads?Scaling up
What’s the largest production environment Kerno has been tested in? What bottlenecks emerged at scale, and how were they solved?Measuring impact
Kerno promises a “64% reduction in customer-facing production incidents” and a “3x increase in successful deployment attempts.” How are those numbers measured and validated?Looking ahead
What would Karim consider a home-run feature or capability for Kerno three years from now?From visibility to autonomy
Will we see a future where observability shifts from passive runtime visibility to autonomous systems that detect and act on issues—powered by eBPF-fed AI models?Startup mindset
If Karim were a fresh graduate looking at the industry, how would he approach finding and validating an idea that could grow into a startup?
🐝 I’ll leave it there—hope you enjoy the conversation.
