About

This repository contains benchmarking tools to compare different RPC endpoints against a list of Starknet JSON RPC methods. Powered by the Criterion benchmarking library.

Prerequisites

• Rust
• In order to generate plots, you should have gnuplot installed

Usage

Configuration file

Edit the configuration file at /config/config.json to add the RPC endpoints and RPC methods of your choice. The format is as follows:

{
    "targets" : [
        {"name": "RPC_NAME", "url": "RPC_URL"},
        ...
    ],
    "methods" : ["METHOD_NAME", ... ],
    "params" : {
        "block": "NUMBER",
        "class_hash": "CLASS_HASH",
        "tx_hash": "TX_HASH"
    }
}

To bench different endpoints, add the pairs "name", "url" to the "targets" array (the "name" field is a custom name of your choice that will appear in the reports). The methods implemented by this library (i.e. the available options for METHOD_NAME above) are:

• starknet_blockNumber
• starknet_getBlockWithTxHashes
• starknet_getBlockWithTxs
• starknet_getStateUpdate
• starknet_getTransactionByHash
• starknet_getTransactionByBlockIdAndIndex
• starknet_getTransactionReceipt
• starknet_getClass
• starknet_getClassHashAt
• starknet_call

Some of these methods need supplementary data like block number, class hash and transaction hash. These are supplied in the "params" field above. Contrary to the other methods, the input parameters for starknet_call are currently hardcoded by some constants.

There are two benching functions. The first one called bench_by_method benches the list of endpoints in "targets" against the list of methods in "methods". The second one called bench_by_block benches the list of endpoints in "targets" on the method "starknet_getStateUpdate" with different block numbers (which are hardcoded in the constant SAMPLE_BLOCK_TAGS).

Run

From the project directory, run

cargo bench

Reports

The output report (including statistics and graphics) will be available at /target/criterion/report/index.html.