I think RAID 1 will have same read performance as RAID 0, Can you please confirm?
There are LOTS of factors involved in RAID performance and even more when it comes to AWS since EBS is network-attached storage…something you normally don’t use with RAID. If you could include some more info about why you think the read performance would be the same, that might help. Here are my supporting ideas for the assertion that RAID0 has better read performance than RAID1…usually.
1) The Intuitive Perspective
It’s generally suggested that RAID0 is faster in both writes and reads….more so in writes because of the striping. But faster reads are usually seen because RAID0 is the sum of the bandwidth across the multiple devices where as RAID1 is limited to the bandwidth of any single device. When we’re network-attached like EBS, bandwidth is important.
2) The AWS Documentation
The below page has some good info on the "official" perspective of AWS on RAID. It indicates the bandwidth gain with RAID0 and the increased performance.
3) My Tests
I was curious on this too so I setup a little test. I spun up a T3.micro EBS-optimized and two 100G EBS SSD NVMe volumes. Setup software RAID using mdadm then used hdparm and dd to benchmark the reads and writes for the RAID volumes.
RAID0 Reads: 224 MB/sec
RAID1 Reads: 170 MB/sec
RAID0 Writes: 246 MB/sec
RAID1 Writes: 135 MB/sec
If you have any material or data to suggest otherwise, please add it to the conversation. Also, if you care to try an experiment of your own, I think it would be a good learning journey too.
I thought AWS Does Support RAID 10 ? 1+0 doesn’t it?
Since the instructor did not mention RAID 10 as an option which seems to be the best of both worlds, I was also in the impression that you could use it in AWS now I’m not sure.
Thanks for the assistance.