Back
Loading views...84_Bare_Metal_Barrelfish_Multi-Kernel_Research_OS
← Back to Index
DEPLOYMENT 84 — Bare Metal → Barrelfish → Multi-Kernel Research OS
Deployment Stack Overlay
┌──────────────────────────────────────────────────┐
│ Workload (App/Process) │
├──────────────────────────────────────────────────┤
│ Multi-Kernel Research OS │
├──────────────────────────────────────────────────┤
│ Barrelfish │
├──────────────────────────────────────────────────┤
│ Bare Metal │
└──────────────────────────────────────────────────┘
Detailed System Infrastructure
┌─────────────────────────────────────────────────────────────────┐
│ RESEARCH APPLICATIONS │
│ ┌──────────────────────────────────────────────────────────┐ │
│ │ Research workloads (custom Barrelfish apps) │ │
│ │ Network stack experiments │ OS research │ HPC protos │ │
│ └──────────────────────────────────────────────────────────┘ │
└────────────────────────────┬────────────────────────────────────┘
│
┌────────────────────────────▼────────────────────────────────────┐
│ BARRELFISH (Multi-Kernel OS) │
│ ┌──────────────────────────────────────────────────────────┐ │
│ │ One CPU Core = One CPU Driver (like a separate kernel) │ │
│ │ ┌────────────────────────────────────────────────────┐ │ │
│ │ │ Core 0: CPU driver 0 (manages core 0 resources) │ │ │
│ │ │ Core 1: CPU driver 1 (manages core 1 resources) │ │ │
│ │ │ Core 2: CPU driver 2 ... │ │ │
│ │ │ Core N: CPU driver N │ │ │
│ │ │ → Each core runs its own kernel-like driver │ │ │
│ │ │ → No shared memory between cores (message passing)│ │ │
│ │ │ → System Knowledge Base (SKB) — distributed fact │ │ │
│ │ │ database for system configuration │ │ │
│ │ └────────────────────────────────────────────────────┘ │ │
│ │ Inter-core communication: URPC (userspace RPC, lockless)│ │
│ └──────────────────────────────────────────────────────────┘ │
└────────────────────────────┬────────────────────────────────────┘
│
┌────────────────────────────▼────────────────────────────────────┐
│ BARE METAL │
│ Many-core x86/ARM (designed for 1000+ cores) │
│ NUMA topology │ Large RAM │ NIC │
└─────────────────────────────────────────────────────────────────┘
Technical Implementation Details
- Management: Orchestrated via native CLI or high-level API controllers.
Use Case
Academic OS research, particularly for heterogeneous many-core systems. ETH Zurich + Microsoft Research project.
Pros / Cons
- Research-only — not for production
- Groundbreaking architecture ideas (multi-kernel model)