Unikernels for Serverless Computing
summary
They build a tiny hypervisor called KVMM hypervisor that moves the VMM into the Linux kernel to cut mode switches, run unikernel images extremely fast, and reduce attack surface; they integrate with kata-containers and OpenFaaS by packaging unikernel binaries inside container images then unbundling and spawning them via a custom runtime (uranc); they add hardware acceleration using vAccel/vxlrt as a multiplexer for FPGA/CPU plugins. Projects are work‑in‑progress but vxl is runnable and claims much faster boot and network I/O than container VMs.
AI-written summary · about
chapters (7)
- 0:00Introduction and challenges
- 2:29Proposed unikernel framework
- 4:13Hypervisor architecture
- 5:53Understanding unikernels
- 8:08Hardware and orchestration
- 10:46Project summary and status
- 12:42Q&A session
related talks
34mLook ma, no OS! Unikernels and their applications
38mUnikernels: The Next Evolution in Cloud Infrastructure
44mUnikernels and Hyper Elastic Clouds
24mLightweight VMs for serverless and containers
20mFirecracker: Lightweight Virtualization for Serverless Applications
27mServerless MicroVMs for Containerized Applications
17mUnikernel based deployments
24mUsing Unikernels to deploy real-world apps