You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Looking at the following code in benchmarks/twitter-finagle/src/main/scala/org/renaissance/twitter/finagle/FinagleHttp.scala it seems to me that val response ... should be moved to the inner loop as in the current setup we sent one request per client and repeatedly read its result. But the naming suggests that there should be NUM_REQUESTS requests sent.
finagle-http is now sending requests in parallel. Also, the measured
part of the benchmark now excludes starting of the threads (there is a
barrier to start all the work at once).
Note that the parameters (number of clients and number of requests) were
changed to roughly correspond to the previous performance (same order of
magnitude).
Also note that this commit changes the performance of the finagle-http
benchmark drastically.
For references, here are the issues summarized:
* Issue 148: single requests was repeatedly waited for instead of
sending the whole request again.
* Issue 149: no parallel requests were actually executed.
finagle-http is now sending requests in parallel. Also, the measured
part of the benchmark now excludes starting of the threads
(there is a barrier to start all the work at once).
Note that the parameters (number of clients and number of requests)
were changed to roughly correspond to the previous performance
(same order of magnitude).
Also note that this commit changes the performance of the
finagle-http benchmark drastically.
This closes#148, #149 and #161.
Looking at the following code in
benchmarks/twitter-finagle/src/main/scala/org/renaissance/twitter/finagle/FinagleHttp.scala
it seems to me thatval response ...
should be moved to the inner loop as in the current setup we sent one request per client and repeatedly read its result. But the naming suggests that there should beNUM_REQUESTS
requests sent.Is this a bug or intentional behaviour?
The text was updated successfully, but these errors were encountered: