now tcp
Concurrency Level: 16
Time taken for tests: 22.963 seconds
Complete requests: 1048576
Failed requests: 0
Total transferred: 1106755384 bytes
Requests per second: 45663.49 [#/sec]
Transfer rate: 47067.47 [Kbytes/sec]
vs normal
Concurrency Level: 16
Time taken for tests: 22.286 seconds
Complete requests: 1048576
Failed requests: 0
Total transferred: 1106754345 bytes
Requests per second: 47050.30 [#/sec]
Transfer rate: 48496.88 [Kbytes/sec]
Performance not so good. Could need some optimization.
Concurrency Level: 16
Time taken for tests: 33.575 seconds
Complete requests: 1048576
Failed requests: 0
Total transferred: 1106753375 bytes
Requests per second: 31230.86 [#/sec]
Transfer rate: 32191.03 [Kbytes/sec]
vs normal http
Concurrency Level: 16
Time taken for tests: 24.829 seconds
Complete requests: 1048576
Failed requests: 0
Total transferred: 1106761259 bytes
Requests per second: 42231.10 [#/sec]
Transfer rate: 43529.78 [Kbytes/sec]
same as happens on the s3 API on delete bucket,
to reduce the IO requests on the system, you can just clear recursively folders without flagging them in volume server (they will be removed on delete bucket/drop collection)
this will remove the user ability to retrieve data under these folders