[Bizgres-general] A bitmap index access method is available in bizgres CVS tip

[Mark Kirkwood]

Mark Kirkwood wrote:

> Looking at the original setup again (the 3 indexes), a quick display of 
> the pagesizes of the various relations shows:
> 
> Bizgres current:
> bitmap=# select relname,relpages from pg_class
>          where relname like 'bitmaptest%'
>          or relname like 'bm_internal_t2200bitmaptest%';
>               relname               | relpages
> ------------------------------------+----------
>  bitmaptest                         |    93458
>  bitmaptest_id                      |    21899
>  bitmaptest_val0                    |    19363
>  bitmaptest_val1                    |    19360
>  bitmaptest_val2                    |    70125
>  bm_internal_t2200bitmaptest_val0_0 |        1
>  bm_internal_t2200bitmaptest_val1_0 |        1
>  bm_internal_t2200bitmaptest_val2_0 |        1
> (8 rows)
> 
> 
> Postgres 8.1:
> bitmap=# select relname,relpages from pg_class
>          where relname like 'bitmaptest%'
> bitmap-# ;
>      relname                        | relpages
> ------------------------------------+----------
>  bitmaptest                         |    88496
>  bitmaptest_id                      |    21899
>  bitmaptest_val0                    |    21899
>  bitmaptest_val1                    |    21899
>  bitmaptest_val2                    |    21899
> (5 rows)
> 
> 
> That (bitmapped) bitmaptest_val2 is relatively big at ~600Mb, so I'm 
> wondering if the 100 distinct values is just at the point where the 
> bitmap is getting inflated. Thoughts?
> 
> Cheers
> 
> Mark
> 
> 

One other thing comes to mind - apologies for the drip-feed of the 
thoughts on this, but its how they are coming to me as well :-)

In fact *all* the (bitmapped) indexes seem to have a fairly large 
on-disk footprint - about the same as btree. Is that expected?

I have not looked specifically at how this has been implemented (am 
going to tomorrow), But thinking aloud (so to speak), concerning my 
(possibly completely wrong - apologies if so) theoretical understanding 
of how bitmap indexes work:

10 value bitmap for 10000000 rows requires 10*10000000/8 = 12500000 (~12Mb).

This is quite small (whereas the 10 value bitmap indexes are ~150Mb).

To actually use the bitmap you need a map between the bitmap "array" 
position and blocknumber,itemid (or similar) of the heap relation. I am 
guessing that is what your "bm_internal_t*" guys do(?).  Is this where 
the extra space is being used? (the on-disk bm_* are very small, so I 
guess it is not).

Cheers in confusion :-)

Mark