96TB Scalable Informatics JackRabbit JR5 unit, shipping out to a customer today (or early tomorrow). These are single thread, single process, single file writes. Taking it out to the track and cracking the throttle, wide open.
[root@jr5-2 ~]# fio sw.fio ... Run status group 0 (all jobs): WRITE: io=65028MB, aggrb=3801.1MB/s, minb=3893.2MB/s, maxb=3893.2MB/s, mint=17104msec, maxt=17104msec [root@jr5-2 ~]# fio sr.fio ... Run status group 0 (all jobs): READ: io=65028MB, aggrb=3257.2MB/s, minb=3335.3MB/s, maxb=3335.3MB/s, mint=19965msec, maxt=19965msec
and the 1TB run
[root@jr5-2 ~]# fio sw-1TB.fio ... Run status group 0 (all jobs): WRITE: io=999.78GB, aggrb=3705.5MB/s, minb=3794.4MB/s, maxb=3794.4MB/s, mint=276285msec, maxt=276285msec [root@jr5-2 ~]# fio sr-1TB.fio ... Run status group 0 (all jobs): READ: io=999.78GB, aggrb=3663.7MB/s, minb=3750.1MB/s, maxb=3750.1MB/s, mint=279484msec, maxt=279484msec
Whats interesting to me, apart from the raw performance for a single machine reading and writing a single uncached bit of data … is that some of our competitors are selling storage clusters with less than one half of this performance, in aggregate, with single read/write speeds measured under 1/20th of this, at much higher price points.
Now imagine a siCluster built out of these, with an Infiniband network (preferably one that didn’t run at 35% of its rated speed 🙁 ). 8 of these per rack could get you some pretty serious bandwidth. And with the 3 TB drives, we’d be at 1.1PB/rack, with roughly 20+ GB/s sustained (yeah, I know its more …)