The problem/use-case that the feature addresses
Currently, the data size in the redis-benchmark is a fixed number, but for most scenarios/users, we would like to benchmark with different data sizes to simulate the real user scenario.
Description of the feature
We can specify the data size range like -d 128-4096. I would like to submit a PR if it makes sense to you.
Comment From: enjoy-binbin
makes sense to me, @filipecosta90 WDYT?
Comment From: bugwz
@git-hulk hi，I have some questions about your idea： 1. If the data range is small during load testing, does it make sense to perform load testing with different data sizes? 2. If the data range is large during load testing, is it necessary to control the proportion of operations with different data sizes? If the proportion of operations with different data sizes cannot be guaranteed, then does this load testing have any meaning?
Comment From: madolson
I'm against this. I think the purpose of this tool is to support generating a fix predictable workload. I would be okay if it supports different "ratios" of requests, (Say 80% of 100byte commands and 20% of 1kb commands).