Maya Spike

[0xean]

Questions:

1. Self hosting costs
2. web implementation (how much can we simply reuse of existing TC implementation)

Potentially unblocks additional chains we could support that come with all the typical questions (wallet support, node and blockbook needs, etc)

[kaladinlight]

Current EKS Setup:

• 4 node groups with 3 availability zones per group:

  • r7a.4xlarge (69% average spot instance discount): ~$830/node group/month
  • r7i.4xlarge (62% average spot instance discount): ~$880/node group/month
  • r7iz.4xlarge (66% average spot instance discount): ~$1,110/node group/month
  • r6a.4xlarge (69% average spot instance discount): ~$615/node group/month
  • r6i.4xlarge (61% average spot instance discount): ~$860/node group/month
  • r6in.4xlarge (57% average spot instance discount): ~$855/node group/month
  • r6idn.4xlarge (57% average spot instance discount): ~$1475/node group/month

• EBS gp3 volumes provisioned with 3000 IOPS @ 125/MBps throughput: ~$80/1TB/month

Cost

• Mayanode + Midgard: ~2TB
  • 3 totaal replicas (1 dev, 2 prod)

Estimated Total Cost: ~$500/mo (this assumes we have no need to scale up our node groups which should be the case)

Chain Support

• Arbitrum (free)
• Bitcoin (redundant)
• Dash (minimal lift)
• Ethereum (redundant)
• Kuji (cosmos-sdk based med/high lift)
• Thorchain (free)
• Radix (unknown/very hight lift)

Operational

• Creation of mayachain multisig
• Creation of mayaname for collecting affiliate fees attached to the multisig
• Configuration of any relevant revenue dashboards

Infrastructure

• Deployment of a Mayanode and associated Midgard instances should be nearly identical to Thorchain
  • Public Mayanode snapshots avaialable for bootstrapping the node
  • Midgard blockstore available for bootstrapping
  • Uknown amount of time to sync from bootstrapped state
• Create DNS records for new maya endpoints

Unchained

• Create new coinstack (should largely be duplicate of thorchain)
  • Conflicting versions of cosmos-sdk will most likely be problematic as we have seen in the past resulting in the need for another fork branch (easy), or another stab and breaking out the golang coinstacks (ideas, but no confirmed solution)
• We should ideally use the mayachain dependency for the maya coinstack even though it is largely the same as thorchain, but important for any future divergence between the chains
• Caution regarding any hardcoded thorchain variables (ex: 0.02 RUNE) for any code reuse

HDWallet

• Ledger uses thorchain app for mayachain actions
• Protobuf types match thorchain, amino types have different namespace thorchain vs mayachain, but should pass through all the same.
• Option to abstract coinname from thorchain for reuse or just dupe which makes sense enough imo. Can make the final call when diving in and ensuring no unknown edge case scenarios.

Web

• Reuse/Dupe/Abstract:
  • Swaps/Streaming Swaps/Longtail Swaps endpoints and logic
  • LP Deposit/Withdraw endpoints and logic (do we want to support?)
  • Memo construction and validation
  • Mimir and halted checks
  • Transaction parsing
  • Chain Adapter
• Need:
  • Maya specific longtail tokens and tradable asset map
  • Caip adapters/constants/utils
  • CSPs/Plugins/chain specific boilerplate
  • serializeTxIndex thorchain special case
  • Seeing as memos match, is there any way to distinguish thorchain from mayachain based on on chain data (tx parsing)
  • Be cognizant of any hardcode values for thorchain logic (ex. asset, network fees, etc)

Other

• No fiat on ramps

[0xApotheosis]

Also check if there are any SaaS providers for a Maya node.