Изменения

New NVMes are Micron 9300 of maximum possible capacity — 12.8 TB. Each of these delivers even more write iopsthan 9200’s: 310k instead of 260k. Everything else remains the same.

The new write performance result for 100 RBD clients is 477029 iops (3636 % more than in the previous test). Remember that it’s still only 4770 iops per a client, though. For 10 RBD clients get the result is better: 294000 iops in their case, that’s which stands for 29400 iops per a client, which is of course better.

* tuned rocksdb: <tt>bluestore_rocksdb_options = compression=kNoCompression,max_write_buffer_number=64,min_write_buffer_number_to_merge=32,recycle_log_file_num=64,compaction_style=kCompactionStyleLevel,<br />write_buffer_size=4MB,target_file_size_base=4MB,max_background_compactions=64,level0_file_num_compaction_trigger=64,level0_slowdown_writes_trigger=128,<br />level0_stop_writes_trigger=256,max_bytes_for_level_base=6GB,compaction_threads=32,flusher_threads=8,compaction_readahead_size=2MB</tt> -  — this divides into:** The main part is probably 64x32x4 MB memtables setting (number x merge x size) instead of default 4x1x256 MB. The effect of this change isn’t really clear for me. It may slightly reduce CPU load because sorting a big memtable is slower than sorting a small one. However, 32x4 compactions aren’t probably that much faster than 1x256.

* they already had debug objecter = 0/0 and the rest of debugsdebug levels set to zero.