CODEWITHSUNDEEP
jmeterwrk
Ganesh S
Ganesh S

Reputation: 431

Difference in throughput between jmeter and apache wrk for same application

I am hitting two straightforward application URLs from jmeter using 100 threads via HTTP requests and using response assertion (200 OK). No cookie handler or nothing else at all used in jmeter. Then using the same machines, I use apache wrk tool (developer asked so) with parameters set to -c100 and -t100. I get throughput difference. The throughput of my application is almost 60 to 80% shown higher via apache wrk compared to jmeter. How do I troubleshoot the difference? Developer is sticking to wrk but I want to resolve the difference. I can provide more info if required but I am not using anything special in jmeter as such.

Upvotes: 1

Views: 1394

Answers (1)

Sasha
Sasha

Reputation: 1411

From my experience wrk is faster, it can deliver as much as up to million requests from single node, quite beefy one though (32 cores with 2 network cards), and it's not apache BTW (while jmeter is). -t100 is probably overkill since your test machine probably has not that much cores. And regarding both throughput and latencies there is wrk2 tool which gives more correct results for general purpose web services (where the load is not coordinated)

Upvotes: 1

Related Questions