Should NAS and SSD give same write performance?

SWB are flushed asynchronously to Block Device as per this post:

We had done “write operation” performance test on namespace1 with NAS storage(Azure) and namespace2 with SSD(local disk) storage(Azure) with same load(sequentially)

We got write latency of 2-3ms on SSD and write latency of 4.5-5ms on NAS.

Can you please explain difference between two ? NAS should be slower than SSD and hence “Device Overload” exception in case of NAS is understandable.

But since data is written to NAS and SSD asynchronously, shouldn’t there write latencies be same ? Can you please explain what part of write transaction flow/algo I am missing that explains write latency difference of few milliseconds between NAS write and SSD write (or write latency for both expected to be same)?

When updating a record, Aerospike must read the record. If not data-in-memory or not recently written, that read will be from disk.

If you are intending to replace the existing record with the new write, you will need to use the ‘replace’ write policy.

This topic was automatically closed 6 days after the last reply. New replies are no longer allowed.

© 2015 Copyright Aerospike, Inc. | All rights reserved. Creators of the Aerospike Database.