Advantage of 1.8TB SAS shelves compared to 15.3TB SSD shelf?

[Maurice-FreeOnes]

I was wondering is there an ADVANTAGE of having 4 shelves full with 1.8 TB SAS disks compared to 1 shelves with 15.3 TB SSD disks?

I can imagine about the advantages of the 15.3TB SSD shelf, as less power costs, less broken disks, more space left  in the cabinet. And I assume faster data throughput?

So I am wondering are there also disadvantages other than investing in new disks again?

Regards, Maurice

[chamfer · Accepted solution]

One advantage would be that with more than one shelf you could have shelf redundancy through the use of mirrored aggregates, but this is an edge use case.   https://docs.netapp.com/us-en/ontap/disks-aggregates/mirrored-unmirrored-aggregates-concept.html#how-mirrored-local-tiers-aggregates-work

There will be an orders of magnitude performance increase when replacing 96x 1.8TB SAS HDD with 24x 15.3TB SSD.  IOPS for example for a HDD is around 100-150 per drive, but a SSD has > 10,000 IOPS per drive.  Throughput for a HDD is around 100-150MBps, but a SSD would be around 500MBps.

Talk to your NetApp rep and they should be able to give you a performance comparison between your current and proposed configurations.

[MaGr]

the mirrored aggregates are a Metrocluster feature.

[SpindleNinja]

Outside of the edge case of mirrored aggrs on the same system- not really IMHO.  

You'd have more capacity with the 15s - (114TiB vs 264TiB) 
Even with all 96 spindles in a single aggr you wouldn't come close to the throughput of the SSDs.  

I'm curious as to why you're asking this question.  

[Maurice-FreeOnes]

I was aksing this question mostly because in the past somebody told me that big SSD's can slow things down. However from the answers this doesn't seem to be the case.

[SpindleNinja]

Larger drives have slower rebuild times.  

[MaGr]

Yes, in case of NL-SAS.

That's why they invented Raid-TEC. Takes just some days to rebuild 10TB a disk.

[MaGr]

Hey,

If you have experienced SSD speed you never want that spinning rust again 🙂

Your disk backend will be never again the bottleneck. 

With a AFF you have more efficiency possibilities to save space.