Why is performance low when using rados gateway in ceph?

Question

Ceph

Colleagues, good day to all.
Set up a cluster:
cluster:
id: 68d893db-be6f-4b52-ba62-b0e99f492e65
health: HEALTH_WARN
1 pools have many more objects per pg than average
services:
mon: 3 daemons, quorum mn-node2,mn-node3,mn-node4
mgr: mn-node2(active), standbys: mn-node3, mn-node4
osd: 15 osds: 15 up, 15 in
rgw: 1 daemon active
data:
pools: 9 pools, 816 pgs
objects: 78.91 k objects, 21 GiB
usage: 89 GiB used, 14 TiB / 14 TiB avail
pgs: 816 active+clean
When testing cluster performance, data was uploaded using rados gateway and rbd.
Below are the statistics:
RADOS GATEWAY
1000 objects of size 10 have been inserted in 68979 millis
1000 objects of size 100 have been inserted in 8229 millis 1000
objects of size been inserted in 28486 millis 1000 objects of size 1000000 have been inserted in 152592 millis RBD 1000 files of size 10 have been inserted in 90 millis 1000 files of size 100 have been inserted in 62 millis 1000 files of size 1000 have been inserted in 61 millis 1000 files of size 10000 have been inserted in 82 millis 1000 files of size 100000 have been inserted in 249 millis
1000 files of size 1000000 have been inserted in 1626 millis
Help me figure out why the speed is 100 times different.

Answer 1

Thanks to all. Figured it out myself! It turns out that the problem is in the code that uploaded the data, it needs to be uploaded to several threads.