AIX > Administrator > Performance

AIX Flash Cache Statistics

Flash Cache

The hit rate is the percentage of read operations that were full read hits. i.e. where all the data requested was in the cache. In this report we see that in the last 60 seconds 98.5 percent of read requests were found in the read cache and the average in the past hour was 96.46 percent.

The partial read hit rate is the percentage of read operations where some portion of the read request was in the cache. Since our hit rate is so high it is not surprising that the partial rate is 0 percent.

Promotes are the percentage of read operations that were issued to the SAN as part of the promote into the cache.

The next two sections include the I/O statistics for the cache pool (DAS) and the SAN disks.

DAS I/O Stats
Avg. data read per second      |      10142010956 |      10097186003 |
Avg. read request size         |           606720 |           609280 |
Avg. read latency (usec)       |             2169 |             2339 |
Avg. data written per second   |                0 |                0 |
Avg. write request size        |                0 |                0 |
Avg. write latency (usec)      |                0 |                0 |
SAN I/O Stats
Avg. data read per second      |            46011 |         14238246 |
Avg. read request size         |            11776 |          1010176 |
Avg. read latency (usec)       |            67412 |            17428 |
Avg. data written per second   |                0 |                0 |
Avg. write request size        |                0 |                0 |
Avg. write latency (usec)      |                0 |                0 |

In the final section, there is a report for each source hdisk on the SAN. As with the global cache operations report there is a cache operations section that reports hit rates, etc for this hdisk, and there is also a LUN I/O statistics section that contains information such as data read per second for this hdisk. When monitoring flash cache performance, it is also useful to look at nmon to monitor the cmpool0 volume group. The VGBUSY, VGREAD and VGWRITE tabs in the nmon analyser report will provide very useful data.


For the right workload, flash cache can make a significant performance difference. We use a combination of real flash for part of our workload and the flash cache to front end the other part of the workload. This combination has made an enormous difference to performance and to the time it takes jobs to run, especially jobs with large sorts or that process large amounts of data. Over time I am hoping that better documentation will come out for the performance monitoring tools, but in the meantime the combination of nmon, pfcras and cache_mgt monitor get should be able to provide you with a strong indicator as to whether flash cache is beneficial for your workload or not.

Jaqui Lynch is an independent consultant, focusing on enterprise architecture, performance and delivery on Power Systems with AIX and Linux.

Like what you just read? To receive technical tips and articles directly in your inbox twice per month, sign up for the EXTRA e-newsletter here.

comments powered by Disqus



2018 Solutions Edition

A Comprehensive Online Buyer's Guide to Solutions, Services and Education.

Achieving a Resilient Data Center

Implement these techniques to improve data-center resiliency.


AIO: The Fast Path to Great Performance

AIX Enhancements -- Workload Partitioning

The most exciting POWER6 enhancement, live partition mobility, allows one to migrate a running LPAR to another physical box and is designed to move running partitions from one POWER6 processor-based server to another without any application downtime whatsoever.

IBM Systems Magazine Subscribe Box Read Now Link Subscribe Now Link iPad App Google Play Store
IBMi News Sign Up Today! Past News Letters