This post is more than 5 years old
29 Posts
0
708
April 28th, 2008 15:00
Which metrics are reliable when using analyzer?
According to issue tracker article emc161309, a lot of the statistics reported are often estimated based on a few actual counters. As an example, it states:
Response time is calculated from utilization, throughput, and average busy queue
length. Average busy queue length is estimated from some of these and a few other counters.
The article goes on to state that some of the counters are available at the metaLUN level while others are available at the component and standard LUN levels. As a result, analyzer tries to estimate values for those statistics and hence often reports inaccurate data.
The fix for this is supposed to be in a future FLARE release after Release 24. My questions are:
1. Has this been fixed in a patch to Release 24 or only in Release 26 (if at all)?
2. If it has not been fixed in Release 24, do the inaccuracies listed in the article hold for Release 24 as well? If not, what are the differences between what the article says and the data for Release 24?
3. Is it possible to get similar data from navicli (Release 24) using getdisk and getlun? I know that getlun for metaLUNs doesn't really offer a lot of information but I don't know whether the data for component and standard LUNs and disks is valid and accurate (at least as accurate as analyzer).
Thanks.
Response time is calculated from utilization, throughput, and average busy queue
length. Average busy queue length is estimated from some of these and a few other counters.
The article goes on to state that some of the counters are available at the metaLUN level while others are available at the component and standard LUN levels. As a result, analyzer tries to estimate values for those statistics and hence often reports inaccurate data.
The fix for this is supposed to be in a future FLARE release after Release 24. My questions are:
1. Has this been fixed in a patch to Release 24 or only in Release 26 (if at all)?
2. If it has not been fixed in Release 24, do the inaccuracies listed in the article hold for Release 24 as well? If not, what are the differences between what the article says and the data for Release 24?
3. Is it possible to get similar data from navicli (Release 24) using getdisk and getlun? I know that getlun for metaLUNs doesn't really offer a lot of information but I don't know whether the data for component and standard LUNs and disks is valid and accurate (at least as accurate as analyzer).
Thanks.
No Events found!