6 ms·
Very interesting use of Firecracker for agent isolation. How do you handle network identity for the agents themselves? Are you using something like mTLS/SPIFF
by mcnamtm1 8mo ago
Very interesting use of Firecracker for agent isolation.
How do you handle network identity for the agents themselves?
Are you using something like mTLS/SPIFFE to identify the workload inside the VM, or is it purely network-policy based at the host level?
Can you explain the "runtime enforcement" and the "enforcing tool proxy?
- difc 8mo agoThanks! Currently network identity is host-based, but in the middle of introducing SPIFFE based on ZTunnel. Should be done in the next couple of days. Runtime enforcement means that any side effects are routed through a proxy (nucleus-tool-proxy) that does realtime checks on permissions and gates the behavior. SPIFFE for MicroVM agents is a compelling idea and I'll update when this is ready.