Previously, JackRabbit JR3 units, with single RAID cards, have been hovering around 750MB/s read and write. This was our second generation unit. First generation units were about 600 MB/s +/- a bit.
The third generation unit is faster.
[root@jr3 ~]# dd if=/dev/zero of=/data/big.file ... 4096+0 records in 4096+0 records out 68719476736 bytes (69 GB) copied, 84.9058 seconds, 809 MB/s [root@jr3 ~]# dd if=/data/big.file of=/dev/null ... 4096+0 records in 4096+0 records out 68719476736 bytes (69 GB) copied, 66.4935 seconds, 1.0 GB/s
Writes are a little faster, reads are significantly faster.
This is borne out in the fio testing.
[root@jr3 ~]# fio streaming2.fio ... Run status group 0 (all jobs): WRITE: io=45,440MiB, aggrb=771MiB/s, minb=771MiB/s, maxb=771MiB/s, mint=58949msec, maxt=58949msec Run status group 1 (all jobs): READ: io=49,152MiB, aggrb=948MiB/s, minb=948MiB/s, maxb=948MiB/s, mint=51875msec, maxt=51875msec Disk stats (read/write): sdc: ios=331776/331778, merge=0/0, ticks=1190980/1542630, in_queue=2733210, util=91.15%
and remembering that 1MiB = 1,048,576 bytes, the write speed in looks to be 808MB/s, and read speed looks to be 994MB/s, consistent with our other streaming test. We were testing at least 2x RAM, so caching isn’t impacting this.
This is a good delta, and this unit gives a good predictive model for that storage cluster we are building, that uses, effectively, 24 of these IO blocks, in a different organization. We are working on borrowing a QDR switch for this … benchmarks should be quite interesting.