Skip to Content

MaxDB on Windows: Allocation Unit Size for Data- and Log-Files

Hello everybody,

the SAP notes

1173395 - FAQ: MaxDB Konfiguration

993848 - Direct I/O Mount Optionen für liveCache/MaxDB

talk about different Storage-Vendors and mount-options for Linux/Unix.

But does anybody know if there are special recommandations on Windows for the NTFS Allocation Unit Size for the MaxDB-Data and Log-Files?

Thank you!

Rossi

Add comment
10|10000 characters needed characters exceeded

  • Follow
  • Get RSS Feed

2 Answers

  • Best Answer
    Apr 07, 2010 at 04:16 PM

    Hi Rossi,

    no for Windows there aren't any special recommendations, you may well just stay with the NTFS default cluster size of 4K.

    In the worst case scenario this means that for a single 8K access two physical IO operations would be necessary.

    But in fact, most often the data volume files are allocated in a few rather large contiguous chunks in the file system.

    Therefore don't bother with that!

    Nevertheless, under certain conditions it might become an advantage to use a 8K cluster size for the file system where you put the data volume files to.

    E.g. if you're using a BI system with MaxDB and therefore using the clustering feature a lot.

    But until now (approx. 6 years at MaxDB support) I never came across a single case where the cluster size for the NTFS file system would have been the root cause for problems.

    regards,

    Lars

    Add comment
    10|10000 characters needed characters exceeded

  • Apr 08, 2010 at 06:31 AM

    Hey Lars,

    thank you very much for your answer!

    Regards, Rossi

    Add comment
    10|10000 characters needed characters exceeded