How to monitor the SSD Cache tier?

[romatlo32]

Hello folks!

I am new to Nimble products and liking it so far.

We are using a CS215 with 4xSSD drives for a cache tier.

I am being asked to verify that a workload (batch file queries) on one of our Oracle database VMs is running in the SSD cache tier.

I see there is a Cache Hit rate monitor, but is there a more detailed method to monitor scheduled workloads and see them use the SSD Cache tier (instead of spinning drive tier) when they are supposed to?

Thanks!

[chris24 · Accepted solution]

Hello and Welcome,

Nimble do not use 'tiering' in the traditional sense, any random IO is populated into cache where it will reside until it ages out.

The only way to see if cache is being used is indeed the cache hit graph, to ensure you get as granular a view make sure the volume is used by only that application.

You can pretty much guarantee that on the second run of the query it will all be in cache.

You can force the IO to be in cache by 'pinning' the volume into cache, however on a CS215 this is unrealistic.

Many thanks,

Chris

[romatlo32]

Thank you Chris!

Just to clarify for me...

A 95% Cache hit rate means that the SSD Cache responded to IO it had seen before 95% of the time (5% missed)?

[chris24]

95% CHIT rate means that 95% of all IO was served by the SSD and 5% from the disks.

[romatlo32]

Thank you!  That is good to know!

I will close this thread.

[rfenton4]

Just a slight correction - Served by SSD or NVDIMM! - It's possible to see cache hits for volumes where caching is turned off (any reads that are resident in NVDIMM tier will be serviced accordingly)