Wind Tunnel scenarios

[ThetaSinner]

This is an epic to cover Wind Tunnel scenarios to be created. It does not need to be complete and can be added to over time

Regression tests for single conductor performance

• App installation speed
• First zome call to a newly installed app
  • To trigger init that succeeds
  • Trigger init which does a network call #108
• Post-init, zome call rate for
  • Calling a no-op function (how quickly can we load, run, return)
  • Database read get performance, (joining proof? get entry on your agent id, parent action of response is the joining proof)
  • Database read chain query performance
  • database read/write
  • Network read
• Local signal send rate
  • Zome fn to emit 10k signals, then measure how long it took and how many signals the client was able to receive
• Validation performance with must_get_agent_activity
• Validation performance for simple structures with must_get_* other than agent_activity

Regression tests for single conductor performance using wasmer_wamr feature

• App installation speed
• First zome call to a newly installed app
  • To trigger init that succeeds
  • Trigger init which does a network call #108
• Post-init, zome call rate for
  • Calling a no-op function (how quickly can we load, run, return)
• Local signal send rate
  • Zome fn to emit 10k signals, then measure how long it took and how many signals the client was able to receive
• Validation performance with must_get_agent_activity
• Validation performance for simple structures with must_get_* other than agent_activity

Network performance

• Time to receive required validation receipts Just measure target redundancy
• Remote call send/receive rate: Multi-Conductor Remote Call Send/Receive Rate #50
• Remote signals send rate Very similar to remote calls for now
• Rate of saturation to reach target redundancy with publishing (fast push)
• Measure how much time is spent on rebuilding the time tree during gossip at the DHT size (history) grows - has to be over time, because that's the partition that should stabilise

Network performance with fault injection

• DHT sync performance with node volatility so that some nodes will come and go during the scenario #66

Countersigning performance

• Two-party countersigning
• Neutral neighborhood countersigning

Validation performance

• Validation rate (in ops/sec)

[ThetaSinner]

Best estimate for this issue is 3 weeks.

[mattyg]

I added a section for "Regression tests for single conductor performance using wasmer_wamr feature" copying the relevant items from "Regression tests for single conductor performance".

Let me know if this issue isn't the best place for that and I should create a standalone issue.