UNSOLVED

infra12341

updated

17 years ago

0

489

September 4th, 2009 01:00

Analyzer Strategy?

What kind of strategy do people use for analyzer?

Is it beneficial to log all of the performance data on an ongoing basis, or should it only be used when problems are encountered? I would like to maintain a complete record of performance information so I can look for trends etc. Is this a bad idea?

I have had it running for a couple of days now, but when I try to retrieve the data I only get the last few hours, and it appears to 'lose' the data between exports, barring these few hours (approx 8). Is this something to do with the 'Periodic archiving' switch?

Keen to hear how others do this...

Thanks,

James
  • RyanP2

    261 Posts

    148

    0

    Posted September 4th, 2009 07:00

    Periodic Archiving is what tells analyzer to create a file when the file is about to wrap, so yes that's why you only get the last x hours. The file only contains 156 data points and depending on the polling rate is when it will wrap.

    I have seen customer keep the archive interval (pooling rate) at 600 seconds so each file contains 26 hours of data. Some even set up a script to collect the nars each night at say midnight so they always cover the same timeframe (collection via cli).

    The 600 second interval (meaning each data point is an average of 10 minutes of data) is good for long trends during the day, and if a problem is happening "all the time" it should show it if a problem exists on the Clariion. For quick issue you need to investigate, you would nee to lower the polling rate to say 120 or 60 to get a better view (files wrap at 5 hours and 2.5 hours respectively).

    -Ryan
  • Allen Ward

    6 Operator

    •

    2124 Posts

    148

    0

    Posted September 4th, 2009 11:00

    We have been using CLARiiON arrays for most of this century (lol) and have found it beneficial to have the performance data gathering all the time. We use the default 600 second interval in order to get a broad view of the last 26 hours. This way, if someone reports a problem we can look back and see what was going on. Otherwise you have to try and catch a repeat of an incident. Not always easy.

    On rare occasions when there are ongoing problems identified we will "crank up" the interval to increase the "resolution". We never leave it this way though. Once we find what we are looking for we always turn it back down to 600 seconds.

    About the only time we turn off the data gathering is during FLARE upgrades. We need to get the SP utilization down as much as possible prior to and during the upgrades.
  • kelleg

    6 Operator

    •

    4537 Posts

    148

    0

    Posted September 8th, 2009 07:00

    When you have Periodic Archiving disabled, the archive will overwrite based on the collection period - if the Archive Interval is set to 600 seconds, then every 26 hours the archive will start to over write - it will keep the data from the time you "create" a new archive back 26 hours.

    If you enable Periodic Archiving, the array will automatically create a new archive when the archive has collected 156 data points.

    Check the On-Line Help - there is a very good section about using Analyzer.

    On PowerLink you can search for Knowledgebase solutions - see articles emc148982 and emc218359.

    glen