Good day all,
I am working on performance tweaking for a database server. I enabled 40GB @ 8K block of vFlash Read Cache. Below is what I am seeing as a result after a few days of usage. The data is SQL DB hosted for a CRM software. Server has a few other DBs as well but small. Are DB servers known to have a low Cache Hit Rate like this? Based on the data below are there suggestions for improving the hit rate? I see that the 40GB vFRC is maxxed out already so would increasing the cache size equate to a better hit rate? The largest DB on this server is only 60GB but the drive its on is 400GB.
[root@svb:~] esxcli storage vflash cache stats get -c vfc-4290317377-
Read:
Cache hit rate (as a percentage): 25
Total cache I/Os: 135730
Mean cache I/O latency (in microseconds): 561
Mean disk I/O latency (in microseconds): 16867
Total I/Os: 544195
Mean IOPS: 153
Max observed IOPS: 584
Mean number of KB per I/O: 530
Max observed number of KB per I/O: 6216
Mean I/O latency (in microseconds): 6531
Max observed I/O latency (in microseconds): 6776
Evict:
Last I/O operation time (in microseconds): 0
Number of I/O blocks in last operation: 0
Mean blocks per I/O operation: 0
Cache usage rate (as a percentage): 99
Total failed SSD I/Os: 0
Total failed disk I/Os: 0
Mean number of cache blocks in use: 1835721
vscsiStatsL: This info was used to determine the block size.
Histogram: IO lengths of Read commands for virtual machine worldGroupID : 242043
6, virtual disk handleID : 8313 (scsi2:0) {
min : 512
max : 524288
mean : 55716
count : 14747
{
1 (<= 512)
0 (<= 1024)
0 (<= 2048)
0 (<= 4095)
81 (<= 4096)
0 (<= 8191)
7854 (<= 8192)
2 (<= 16383)
136 (<= 16384)
749 (<= 32768)
224 (<= 49152)
154 (<= 65535)
4473 (<= 65536)
11 (<= 81920)
182 (<= 131072)
158 (<= 262144)
722 (<= 524288)
0 (> 524288)
}
}