Flash Pool - AWA detailed output - Low cacheable reads with high read hit rate

[pedro_rocha]

Hi,

I am trying to analyze Flash Pool candidacy and optimal cache size for an ONTAP cluster. It is running ONTAP 9.3P13. I followed the instructions on this link.

So far so good, been able to interpret most of its output. But then I tried to see the detailed output using the -t option in the print command. It shows information about the volumes in the aggregate that are the best candidates for being on a Flash Pool aggregate.

Happens that for a specific aggregate, the #1 volume in the list is the following:

Volume #1 VOL_XYZ
Summary of the past 7 intervals
                                                     max
                                                     ------------
Read Throughput (MB/s): 21.545
Write Throughput (MB/s): 12.757
Cacheable Read (%): 0
Cacheable Write (%): 12
Max Projected Cache Size (GiB): 1.856
Projected Read Hit (%): 50
Projected Write Hit (%): 8

So, could someone explain why and how could this volume have 0% of cacheable reads and even so present a Projected Read Hit of 50%?

Regards,

Pedro Rocha.

[paul_stejskal]

Because it is inserted with the writes. If you need specific help,  your SE on your account team can help you review the AWA output as well.

[pedro_rocha]

Hello Paul,

I just wanted to understand that output. There are other volumes with apparently strange values. Take a look:

Volume #3 vol_ZXC
Summary of the past 7 intervals
max
------------
Read Throughput (MB/s): 17.398
Write Throughput (MB/s): 3.266
Cacheable Read (%): 0
Cacheable Write (%): 0
Max Projected Cache Size (GiB): 0.036
Projected Read Hit (%): 12
Projected Write Hit (%): 0

"No cacheable data" but 12% read hit, top 3. How come?

Besides that it would be nice to understand what are the metrics that made up this top 8 rank. Looking at them does not seem to translate the aggregate recommendation.

Pedro