$ git clone <this repo> --recurse-submodules Install docker. $ ./proto-gen.sh $ pip install protobuf==3.14.0 grpcio Install kubo on both IPFS peer host and IPFS canary peer host. Install docker on IPFS peer host, ensure sudo-less docker setup. Ensure password-free login to both IPFS peer host and canary peer host. Initialize main IPFS peer (but not starting daemon) on an instance $ cargo run --bin init-instance -- <INSTANCE DNS> Start a telemetry container along with initialization $ cargo run --bin init-instance -- <INSTANCE DNS> telemetry ---- Recommended evaluation procedure. Stage 1 1.1 Run `dump-providers` with apse instance 1.2 Run `find-providers` with sae instance Cannot run `profile-providers` for now, because dumped provider records are not ready yet. Stage 2 2.1 Run `profile-providers` with sae instance 2.2 Run `find-providers` with apse instance Do not run `profile-providers` concurrently from multiple instances to avoid interfering. Stage 3 3.1 Run `profile-providers` with apse instance It's probably unnecessary to dump providers for multiple times at different locations. The difference in results should only because of temporal effect, not geological one.