This proposal is designed to incentivize a few teams to independently maintain a robust API for public use for the Secret Community. Each of these teams has proven an ability to deploy, respond to issues, and maintain APIs for the network. Each team will be compensated $150 monthly per secret node they offer in their cluster with a cap on incentives at 20 nodes per team.
Currently for teams we have Secret Saturn, Trivium, and Secret LLC.
Combined Cluster Load Balancer
A additional load balancer will be deployed that connects all the nodes from each team.
Reports for these APIs will all look similar to Server Statistics and each team will be responsible for generating these reports. Secret LLC will also be responsible for sharing tutorials with the other API teams for how to generate these reports. Secret LLC will also be responsible for deploying and maintaining the combined load balancer.
Node Counts & Total Costs
With 3 teams this proposal will cost $27,000 (3 teams x 3 months x 3000$) and provide 60 total API nodes to the community.
These numbers are not including the 10% volatility buffer. But the final ask will include it.
I don’t see Keplr mentioned here, but most of the API issues I hear about are related to the current Keplr cluster. Is that what this proposal is meant to solve? Or would this just be a general purpose public API similar to the ones listed on GitHub - scrtlabs/api-registry: A registry of API endpoints for mainnet and testnet?
Same question, i would be more interested in this proposal if an effort is taken to work with SCRT labs to replace the current API they uphold for Keplr to function.
So I’ve had a conversation with POKT Network in the past about similar initiatives. I don’t think people liked the idea of having to buy the POKT token for it at the time. Curious if it’s a conversation that could be brought up again here. It would open up the ability to provide nodes to the API to anyone.
SCRTLabs has released an update (thanks @Cashmaney) which should improve the performance of existing API nodes (see here: Release v1.5.1 · scrtlabs/SecretNetwork · GitHub). This should increase performance with certain requests and helps with the ongoing Keplr performance issues. It does not however change the enclave performance, we simply need more nodes for that.
We still think that the API Prop is therefor necessary despite the extra performance improvements, since:
Existing API nodes can crash (mostly app hash or enclave issues) which can cause delays etc. Then there is also this is kind of self-reenforcing cycle: Under heavy usage API nodes may fall behind, which causes even more requests from users because Dapps don’t work which causes even more API nodes to get slower etc… More nodes will help here.
Have more API Nodes available for teams that cannot afford their own nodes, we make their onboarding into secret easier and also not as cost intensive. Especially with the community efforts to bring in new teams we can ease the transition into Secret Network. For Teams who don’t have their own node runners and are unexperienced this an unnecessary barrier that we can fix.
Have backup API Nodes available 24/7 incase the teams nodes break down or get overloaded. (See below)
Also we would like to answer the question about the API node performance for the Bushi NFT mint and Shade Airdrop (congrats to the teams on the successful start!):
The Bushi NFT mint from an API node side was fine, nothing extraordinary to report, nodes held up fine.
For the Shade airdrop we can report that the Shade team also used public API nodes after some of their own didn’t work. Without the public API nodes the launch wouldn’t have been as smooth as it is now, and some were on the brink of being overloaded as far as we can tell.
After gathering some feedback we got additional parties into the boat:
Consensus.one and Chain of Secrets will also join us on the API Prop. To balance out the additional parties, each node running team will only 16 instead of 20 nodes (exception is CoS).
In summary, we will have these nodes available: 16x Secret LLC, 16x Trivium Network, 16x Secret Saturn, 16x Consensus.one, 8x Chain of Secrets.
The total ask therefore adjusts to (2400$ x 4 teams + 1200$ CoS) x 3 months = 32,400$.