[Load Testing] Permissionless Demand Implementation Umbrella Ticket

[Olshansk]

Objective

Ensure there will not be any scalability issues in Shannon.

This is done by:

• Avoiding free loaders (suppliers that don't submit proofs for claims)
• Keeping blocks size small
• Creating incentives & disincentives to submit proofs
• Enabling unlimited number of services & apps

Origin Document

• The developer needs to fully review and understand this document: Probabilistic Proofs Document from Morse
• Reference (if needed) the notes here: Permissionless Demand Load Testing Notion Doc
• It is a collaboration with the effort from @okdas here:
  • [Load Testing] Permissionless Demand Implementation Umbrella Ticket  #758
  • [LoadTesting] Permissionless Demand Load Testing #711
• This block is 2.5MB, which is already causing scalability issues in shannon. We need to start iterating on probabilistic proofs!

Goals

• Understand missing gaps in the implementation of probabilistic proofs on both the supply (relayer) side & consensus (validator) side
• Implement everything necessary to make sure Shannon demand side scales scales
• Identify MUST have TODO for mainnet
• Come up with new ideas to improve the overall design while keeping things simple
• Update the documentation so all readers and ecosystem participants understand how things work
• Keep iterations and PRs small because the scope of this can get out of hand

Deliverables

• Ensure the ProofParams is used and reflected in the RelayMiner. It doesn't seem these are being used:

   type ProofParams interface {
   	GetRelayDifficultyTargetHash() []byte
   	GetProofRequestProbability() float32
   	GetProofRequirementThreshold() uint64
   	GetProofMissingPenalty() *cosmostypes.Coin
   }

  • Is it being used off-chain?
    • If yes -> just add a few comments (already building context)
    • If not -> use it. [SessionManager] Implement off-chain proof params usage #765
    • Along the way, keep eyes out for issues or gaps -> either add TODOs or fix it on the spot

• Increase the tx fee of the SubmitProof msg so it's larger than a simple send

  • Update docs -> call out that this is needed to sybil attacks and state bloat. Why? [Docs] Add operations documentation about proof submission fee #806
    • Why? Claims are small and tokenomics will slash you.
    • Why? Proofs are big, so you have to pay for the space.
  • Look into the cosmos correct way of doing this. Proof-submission-cost-Tx-fee-vs-Spend-coins
    • If it's appropriate -> as a reference, the only other message in potroll today where we charge the signer a custom amount is AddService, it can be used to learn and other understand properitery ways of doing this. Note that the only reason we didn't look into using cosmos tx fees in the past, is because we didn't have time.
    • If not -> use the same pattern as in AddService
      • [Service] Align add service fee param with other params implementation #778
      • [Proof] Proof submission fee param #779
      • Feat/payable proof submission #780

• @Olshansk to build documentation for this. Specifically:

  • Probabilistic proofs in the Shannon. Tokenomics and incentives.
  • All the slashing conditions. Specifically, missing proofs.
  • Design & document feedback loop between Relay Mining & Probabilistic Proofs
  • Research & document whether proof_requirement_threshold needs to be a function of supplier stake. Look into x/tokenomics/keeper/settle_pending_claims.go
  • Penalize / slash the supplier if a proof is required but not available. Note: you need to determine what the slashing amount should be.

• Implement the Relay Mining & Probabilistic Proof multiplier + closed loop systems [Tokenomics] Implement difficulty proportional rewards #840

• Implement the supplier stake <> proof requirement threshold closed loop system

• Implement slashing for missing proofs [Supplier] Initial slashing implementation #795

• Look for TODOs in the code that call out SparseCompactClosestProof

  • Evaluate whether it's possible to use them, and make the switch if so. [SMST] feat: Use compact SMST proofs #823
  • Provide metrics and solutions

• Exploratory: Review all the probabilistic proof parameters and make sure they adjust dynamically based on the volume, similar to relay mining.

Non-goals / Non-deliverables

• The following are requirements of [LoadTesting] Permissionless Demand Load Testing #711 being built by @okdas
  • Build the tooling for these load tests
  • Building observability tools for these tests
  • Gain an understanding into the block size
  • Get visibility
• Diagrams that will be needed but outside of the scope:
  • Line charts showing:
    • Block size (in MB)
    • Num proofs per block
    • Total number of relays in network
  • Double bar chart showing average number of proof per application AND per service
  • Relay mining difficulty per service (already have this but make it more accessible)
    • Same chart needs to show what the multiplier looks like

---

Creator: @Olshansk
Co-Owners: @okdas @red-0ne