Benchmark application rpc endpoints

[ManuGowda]

Description

Application endpoints should be benchmarked and all the endpoints should respond within a standard timeframe (100-500ms, we can also define the timings).

Motivation

• Lisk Service receives lot of timeout errors of certain actions and endpoints on Lisk Core.
• This intern has impact on Lisk Service for
  • Indexing
  • Timeouts at Lisk Desktop which leads to bad user experience.
• Also some of the exchanges and third party applications are using Lisk Core/Lisk Service, these timeouts can be a bottleneck for the users.

Additional Information

• While benchmarking make sure we consider inter-process, intra-process, docker environments into consideration, as Lisk Service runs in docker container and Lisk Core might be running in a docker container or on a host machine
• Consider possibility to configure timeouts, so that one could increase when required
• Endpoints (we also have other actions in modules, take them into consideration too) : https://github.com/LiskHQ/lisk-sdk/blob/c9c94a4fc0faaf72193d9689df946b6a87511887/framework/src/application.ts#L399-L476

[sameersubudhi]

The endpoints most frequently resulting in timeouts at Lisk Service ​are:

• legacyAccount:getUnregisteredAccount (On concurrent requests)
• app:getAccounts (On concurrent requests)
• app:getAccount (On concurrent requests)
• app:getLastBlock (Occasionally)
• app:getForgers (Occasionally)
• app:getTransactionsFromPool (Occasionally)
• app:getTransactionsByIDs (On concurrent requests)
• app:getNodeInfo (Mostly likely when Core is under pressure and others are timing out)

[shuse2]

Closing this issue since this is no longer valid