Post reply

IndexID = 0?

  • March 17, 2005 at 4:48 pm

    #162901

    Ok, I'm confused!

    I have a table called CREDITORS. This Table has a Primary Key with a NONCLUSTERED INDEX on CRED_ID. This means that there is only 1 Index on the Table.

    If I execute this command; DBCC SHOWCONTIG (CREDITORS)

    I get fragmentation info for an IndexID of 0. What is this Index? There is only 1 Index on the Table and it has an IndexID of 2 which happens to be the NONCLUSTERED INDEX in the Primary Key for CRED_ID.

    If I execute this command, I get the correct fragmentation info for the Index CREDITORS_PK which is the NONCLUSTERED INDEX in the Primary Key for CRED_ID.

    Can someone please explain?

    Kindest Regards,

  • March 17, 2005 at 8:35 pm

    #547701

    It's a little buried in BOL for DBCC SHOWCONTIG, but Index ID 0 refers to a heap table, i.e., one without a clustered index.

    IndexIdID of the index. IndexId is 0 for a heap.

    A given table will always have one entry in sysindexes for either IndID = 0 or IndID = 1.  Of course, it can have other entries for every non-clustered index as well.

    Scott Thornburg

  • March 17, 2005 at 9:41 pm

    #547710

    so therefore, there really is 2 Indexes to this table? 1 for the Heap and 1 for the Primary Key(NONCLUSTERED INDEX)?

    Having said that, if the heap Index is fragmented, do you defrag it like you would with a normal Index?

    Kindest Regards,

  • March 17, 2005 at 10:35 pm

    #547714

    It is not accurate to say that the table has two indexes in this case.  Unlike a table with a clustered index (IndID = 1) which is well-ordered based on the index key, a heap table has no logical order.  It is simply a conglomeration of data.  Your table has just one index -- the non-clustered one you created.

    As such, for a heap table there is no concept of defragmenting the table.  The standard methods of DBCC INDEXDEFRAG and such do not work. 

    There can be issues with heap tables, see for example, Kalen Delaney's explanation at http://www.windowsitpro.com/SQLServer/Articles/ArticleID/25198/pg/2/2.html.  My understanding is that the only way to compact a heap table is to either reload it (move data out, truncate, move back in ) or to shrink the database itself.   I may have missed a method or two here, however.

    Scott Thornburg

  • March 17, 2005 at 10:57 pm

    #547715

    I'll have aread of the article however, I'm able to retrieve fragmentation info for the IndexID of 0. I must say, it is very fragmented.

    Kindest Regards,

  • March 18, 2005 at 12:54 am

    #547725

    See if this helps:

    http://www.microsoft.com/technet/prodtechnol/sql/2000/maintain/ss2kidbp.mspx

    --
    Frank Kalis
    Microsoft SQL Server MVP
    Webmaster: http://www.insidesql.org/blogs
    My blog: http://www.insidesql.org/blogs/frankkalis/[/url]

Viewing 6 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic. Login to reply