Re: Performance impact of NULLs and variable length fields

Martijn van Oosterhout wrote:
> Sometime ago somebody asked if it made a difference adn it was suggested
> that the effect was probably marginal. I ran a profiler over postgres doing
> a large query and these are the top 10 functions:
>
>   %   cumulative   self              self     total
>  time   seconds   seconds    calls  ms/call  ms/call  name
>  16.04      0.51     0.51  1676772     0.00     0.00  nocachegetattr
>  11.95      0.89     0.38  1427403     0.00     0.00  heapgettup
>  10.06      1.21     0.32  2955372     0.00     0.00  LockBuffer
>   6.92      1.43     0.22  3406475     0.00     0.00  ExecEvalExpr
>   5.03      1.59     0.16  1617018     0.00     0.00  AllocSetReset
>   4.40      1.73     0.14  1427403     0.00     0.00  heap_getnext
>   4.09      1.86     0.13  1844339     0.00     0.00  ExecEvalVar
>   4.09      1.99     0.13  1441330     0.00     0.00  ReleaseBuffer
>   4.09      2.12     0.13  1427064     0.00     0.00  SeqNext
>   3.46      2.23     0.11  1552338     0.00     0.00  ExecQual

I have added this to the TODO list:

    * Improve caching of attribute offsets when NULLs exist in the row

I have some ideas about how offsets with NULL's could be handled in this
case.

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026