TerraMaster D2-310 2-Bay DAS Performance
We will run our benchmarks using 2x HGST 10TB Hard Drives and 2x Crucial M500 SSD’s using RAID 0 and RAID 1 configurations. Generally, we suggest RAID 1 for redundancy. We are going to run through a number of utilities to see the performance. We are going to first look at the 10TB hard drive performance, then we are going to look at SSD performance.
Here we see the results of our RAID 0 and RAID 1 tests using the HGST 10TB hard drives.
One can see the impact of using the simple RAID versus a more robust, and costly, solution. Read speeds are good but write speeds are less than half of what one can get on 1GbE networking using one of TerraMaster’s NAS units and hard drives. There is a massive delta between hard drive and SSD performance.
ATTO Disk Benchmark HPerformance
ATTO shows almost double performance in RAID 0 at 450mb/s vs 250mb/s in RAID 1, it does drop off at large test loads.
That fall-off is quite notable. It likely indicates an architectural limitation of the embedded RAID processor.
SSDs, on the other hand, are able to perform well and not see the same drop off in performance.
Here are the Parkdale hard drive RAID 0 results:
Here are the Parkdale SSD RAID 0 results:
Here are the Parkdale hard drive RAID 1 results:
Here are the Parkdale SSD RAID 1 results:
Overall, these are good results and about what we would expect from the DAS. When we see similar read or write speeds using SSDs and HDDs, it is usually a limitation of the interface or the DAS.
PCMark 8 Benchmark
Here are the PCMark 8 hard drive RAID 0 results:
Here are the PCMark 8 SSD RAID 0 results:
Here are the PCMark 8 hard drive RAID 1 results:
Here are the PCMark 8 SSD RAID 1 results:
There is fairly clearly a limitation here in terms of random I/O performance, largely due to using hard drives rather than SSDs. This is one of those use cases that SSD vendors will trumpet as a reason to use SSDs over HDDs.
Blackmagic Disk Speed Test
We have also started using BlackMagic Design Disk Speed Test to show those doing video how well storage performs. We show RAID 0 on the left and RAID 1 on the right.
Using hard drives in RAID 0, ProRes 422 HQ at 2160p30 is possible. Using RAID 1 will limit one to 50fps editing. That is a practical limitation of the design.
Here, one can see that ProRes 422 HQ at 2160p60 is possible which is a major upgrade over hard drives in the unit.
Next, we are going to look at key specifications as well as our conclusion.