resp-benchmark

The resp-benchmark tool is a command-line "RESP" benchmark client, comparable to redis-benchmark, and many of the arguments are the same. This is mostly for internal team usage, but is included here for reference.

Example usage:

> dotnet tool install -g RESPite.Benchmark

# basic usage
> resp-benchmark

# 50 clients, pipeline to 100, multiplexed, 1M operations, only test incr, loop 
> resp-benchmark -c 50 -P 100 -n 1000000 +m -t incr -l

Options supported from redis-benchmark

Basic options, for parity:

• -h <hostname> Server hostname (default 127.0.0.1)
• -p <port> Server port (default 6379)
• -c <clients> Number of parallel connections (default 50).
• -n <requests> Total number of requests (default 100000)
• -P <numreq> Pipeline <numreq> requests. Default 1 (no pipeline).
• -l Loop. Run the tests forever
• -q Quiet. Just show query/sec values
• -t <tests> Only run the comma separated list of tests. The test names are the same as the ones produced as output.

Custom options

Additional options specific to this tool:

• +m / -m: enable or disable (default) multiplexing: when enabled clients share a connection, otherwise each client has a separate connection
• --batch / --queue pipelining should using batching (default) or queueing strategy
• --basic : perform basic typical IO operations rather than synthetic benchmarks

Internal options

These exist mostly for Marc's benefit:

• -w <mode> Specify the internal write-mode
• +x / -x: enable or disable (default) cancellation support (irrelevant until later v3 tranche)

Local example

To build and run from source, dotnet run can be used with everything after -- being args to the command:

dotnet run -p:TargetVer=3 -f net10.0 -c Release -- -q +m --batch -n 500000