Automatic Storage Tablespaces (AST): Compare and Contrast to DMS

You may also like...

10 Responses

  1. Rajesh Chandramohan says:

    ALTER TABESPACE REDUCE MAX –> This is available in 9.7 but we identified a memory leak with the above command and its fixed in Fixpack 9. Check this for more information :

    http://www-01.ibm.com/support/docview.wss?uid=swg1IC92740

  2. Jim Reutener says:

    Hello

    We are moving from v8.2 to v9.7 to v10.1. Since SMS & DMS are deprecated, I’ve created all new tablespaces (storage group + automatic storage) and “ADMIN_MOVE_TABLE” the user tables. What recommendations do you have for the SYSCATSPACE?

    Tablespace Configuration:
    Address Id Type Content PageSz ExtentSz Prefetch BufID BufIDDisk FSC NumCntrs MaxStripe LastConsecPg Name
    0x000000023AC7D8E0 0 SMS Any 4096 32 0 1 1 On 1 0 31 SYSCATSPACE

    Thanks,
    Jim

    • mkrafick says:

      I struggled with system Tablespaces too and found no real specific best practice to follow. So I defaulted to what I could find which is “to stripe everything everywhere” as best I can with Automatic Storage (quote from “Best Practices: Database Storage” – by Sachedena, Huras, and Grosman). In my personal scenario this is where I found it OK to accept defaults – I spread the AST over multiple file systems is possible, accepted the 32k extent size because is adequate *most* of the time (check “Setting the table space extent size in virtualized storage environments” in the same Best Practices Article on developerWorks), and kept it in the default bufferpool as everything else was moved into defined Bufferpools.

      • Jim Reutener says:

        In regards to syscatspace, I found the following comment in both the v10.1 & v10.5 Info Centers under the heading “Creating table spaces” …

        Important: …The SMS table space type is not deprecated for catalog and temporary table spaces.

        Important: … The DMS table space type is not deprecated for catalog and temporary table spaces.

  3. Bharani says:

    Hi,

    Kindly let us know how the load operation on a table works when the tablespace is being reduced using ” alter tablespace tbpacename reduce max”.

    Is there any risk of losing data if the load on table fails and the table goes to load-pending state.

  4. Richard Kast says:

    Now that there has been some time for experience, do you have any recommendations for tablespace and db pathing In an environment where storage is on SAN with no visibility to the physical storage, multiple different systems are sharing the SAN ( and may not be isolated) and the LUNs presented to the db are already striped across most of the disks.
    Is there any point in having more than one storage path ?

    • Ember Crooks says:

      I tend to go with two containers now mostly because it tends to allow a bit more flexibility. I don’t think there’s anything wrong with one. I don’t see any reason for more than two, unless you do have some separation of IO as far as caching or physical disk isolation goes. Or if you think you might have to add a container down the road for some reason. I have a database now that fully lives on SSD. They need to add more local SSD, and present it as a separate disk. To add containers on that disk, I have to match the size of the current containers (containers of a tablespace should always be equal in size for very complicated reasons). To add containers, I have to add a minimum of 290 GB for one of the tablespaces. If I had more than one container already, the added container could be smaller.

  1. March 24, 2014

    […] wrote a compare and contrast article between Automatic Storage (AST) and DMS storage types in the past. However, in developing my DB2′s Got Talent Presentation on AST, I […]

Leave a Reply

Your email address will not be published. Required fields are marked *