MettEagle: Costs and Benefits of Implementing Containers on Microkernels
summary
Prototype MettEagle runs on L4Re with only small, three‑digit LOC changes to the base system, adds an in‑memory filesystem and UDP/IP stack, and ports Python 3. On a dual‑socket 64‑core server it shows cold‑start latency comparable to Linux processes and faster than run and Firecracker, scales until coarse‑grain L4 locks hurt parallelism, and yields SEPS serverless latencies within 5–15% overhead of Linux containers while outperforming VM‑based isolation; file compression lags due to an unoptimized RAM FS.
AI-written summary · about
chapters (9)
- 0:00MetEagle and container security
- 1:21Container isolation on monolithic kernels
- 3:49Monolithic kernels drawbacks
- 4:31Microkernels improve container security
- 6:37Containers on microkernels benefits
- 8:21MetEagle implementation
- 10:30MetEagle performance benchmarks
- 13:19Serverless function benchmarks
- 15:05MetEagle costs and benefits
related talks
38mSecure Containers in Embedded Deployments
34mContainer Runtimes... on Lockdown: The Hidden Costs of Multi-tenant Workloads
25mOne Size Fits None: Lightweight Workload-Specialized Confidential Containers
14mYour Containers Aren’t Alone: Demystifying Container Isolation
31mYour Containers Aren’t Alone: Demystifying Container Isolation
25mContainers. Microservices. Cloud. Open Source. Fantasy or Reality for Enterprises?
41mNabla Containers: A New Approach to Container Isolation
25mFrom Cloud To AI: How Kata Containers 4.0 Reinvents the Sandbox for the Agent Era