SEARCH
diglloydTools™
|
|
|
|
|
OWC Mercury Elite Pro Dual: Performance with 36TB RAID-0
Related: hard drive, Other World Computing, OWC Mercury Elite Pro, RAID, RAID-0, SoftRAID, SSD, storage, Thunderbolt, Thunderbolt 3, USB, USB-C, weather events
MPG tested the OWC Mercury Elite Pro Dual USB-C in various configurations using two Western Digital 18TB hard drives, which are the fastest hard drives MPG has ever tested.
This page looks at performance of dual Western Digital 18TB hard drives used for RAID-0 striping in the OWC Mercury Elite Pro Dual USB-C.
Sustained performance over 36TB capacity: hardware RAID-0 vs SoftRAID RAID-0
Test mule was the 2018 Mac mini connected to a dedicated Thunderbolt 3 port/bus. 10 Gbps bandwidth verified.
For over 12 years, DiskTester has been the best means available to ascertain how a drive performs across its capacity. When working with RAID, graphing each hard drive individually can reveal performance irregularities that can cause hiccups in RAID performance when multiple drives are used—problem drives can be swapped out for better ones.
The two Western Digital 18TB hard drives are nearly perfectly matched, as demonstrated—perfect for RAID-0.
This test evaluates the hardware RAID-0 performance to SoftRAID over the 36TB capacity.
The overall performance is fantastic for a dual hard-drive RAID-0. RAID-0 performance averages 442MB/sec for writes and 446 MB/sec for reads, identical for hardware RAID-0 vs SoftRAID, on average. That’s an average over the entire 36TB of capacity.
Average Write speed, hardware RAID-0: 442 MB/sec
Average Read speed, hardware RAID-0: 446 MB/sec
Average Write speed, SoftRAID RAID-0: 442 MB/sec
Average Read speed, SoftRAID RAID-0: 446 MB/sec
Those looking to keep performance high might choose to limit capacity usage to the first 80% or so of the capacity, which would guarantee a minimum performance of 360 MB/sec.
Vertical scale MB/sec. Horizontal scale percentage of volume capacity. Red/orange for writes, green/blue for reads.
Data using fill-volume command of diglloydTools Disktester.
NOTE: the spike to peak performance at the very end has to do with file system mapping of blocks (it is not really at the end of the capacity). This file system behavior cannot be controlled. Visualize it inserted at the beginning (at left).
