Re: ESS/Shark performance problems? - Storage

This is a discussion on Re: ESS/Shark performance problems? - Storage ; Mike, I also admin two ESS ("Sharks"). Usually, your storage vendor or an IBM rep will be able to perform a comprehensive performance analysis. Just curious -- how are you collecting these stats? Are you using IBM's ESS Expert 2.1.1? ...

+ Reply to Thread
Results 1 to 2 of 2

Thread: Re: ESS/Shark performance problems?

  1. Re: ESS/Shark performance problems?

    Mike,

    I also admin two ESS ("Sharks"). Usually, your storage vendor or an
    IBM rep will be able to perform a comprehensive performance analysis.

    Just curious -- how are you collecting these stats? Are you using
    IBM's ESS Expert 2.1.1?


    Brian J. Mitchell
    Systems Administrator
    TRX
    6 West Druid Hills Drive
    Atlanta, GA 30329-2158 USA
    http://www.trx.com

    email: brian.mitchell @ trx.com
    office: +1 404 327 7238
    mobile: +1 404 725 3128


  2. Re: ESS/Shark performance problems?

    On 2005-11-27, brian.mitchell@trx.com wrote:
    > Mike,
    >
    > I also admin two ESS ("Sharks"). Usually, your storage vendor or an
    > IBM rep will be able to perform a comprehensive performance analysis.
    >
    > Just curious -- how are you collecting these stats? Are you using
    > IBM's ESS Expert 2.1.1?


    We've engaged IBM almost to a 'critsit' (critical situation?) due to
    the bad performance the intel boxes are getting seemingly from the
    shark. I'm loaded an old F30 with AIX 5.1 ML 9 and the ESSCLI program.
    I have a shell script started from cron(8) to pull statistics from
    the shark. I have been sent a program in python by IBM that reads
    the performance data and generates a CSV file for loading into a
    spreadsheet also from IBM.

    I see the numbers they are now looking at for the performance issues.
    It seems like the mainframe is taking most of the I/Os, the longest
    delays, etc., and is having no performance issues. This coming week
    I think IBM has engaged Cisco to look at the SAN switches that only
    intel is using. The mainframe accesses the shark through 'SCON'
    connections. The AIX boxes use SCSI connections. Only the intel boxes
    are seeing the problem and only the intel boxes use the SAN fabric.

    It could be SDD drivers, or QLogic drivers, though all the drivers
    (intel and AIX) are at the highest levels.

    Mike

+ Reply to Thread