Day good friends.
Need your help on deploy'th cluster in Ceph.
There are several nod on Centos 7. According to the documentation, I deployed Ceph cluster, and RadosGW.
But here's the thing: when you test and determine the performance indicators of the cluster, it was found that there are queries which, unlike most hang in RadosGW of the order of a few seconds. In the end there is not very positive indicators:
Put 30-40/s of objects of different Rahmonov from 4KB to 16 MB.
The nodes are all equal and have the following characteristics:
2 x CPU Xeon e5 2650
128 GB DDR4 RAM
36 OSD (2.5 TB per OSD)
Network 2 x 10 GB in "mode=balance-xor"
SSD, 800 Gb