Re: Mapping a database completly into Memory

You make an interesting distinction that read/write needs more shared
memory.  I think this is because if you want to reused a read-only
shared buffer, you can just throw away the contents, while a dirty
buffer requires you to write it into the kernel before you can use it.

---------------------------------------------------------------------------

Vivek Khera wrote:
> >>>>> "TL" == Tom Lane <tgl@xxxxxxxxxxxxx> writes:
> 
> TL> Franco Bruno Borghesi <franco@xxxxxxxxxxxxxxxxxxx> writes:
> >> wouldn't also increasing shared_buffers to 64 or 128 MB be a good
> >> performance improvement? This way, pages belonging to heavily used
> >> indexes would be already cached by the database itself.
> 
> TL> Not necessarily.  The trouble with large shared_buffers settings is you
> TL> end up with lots of pages being doubly cached (both in PG's buffers and
> 
> I think if you do a lot of inserting/updating to your table, then more
> SHM is better (and very high fsm settings), since you defer pushing
> out the dirty pages to the disk.  For read-mostly, I agree that
> letting the OS do the caching is a better way.
> 
> 
> -- 
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
> Vivek Khera, Ph.D.                Khera Communications, Inc.
> Internet: khera@xxxxxxxxxxx       Rockville, MD       +1-240-453-8497
> AIM: vivekkhera Y!: vivek_khera   http://www.khera.org/~vivek/
> 
> ---------------------------(end of broadcast)---------------------------
> TIP 3: if posting/reading through Usenet, please send an appropriate
>       subscribe-nomail command to majordomo@xxxxxxxxxxxxxx so that your
>       message can get through to the mailing list cleanly
> 

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@xxxxxxxxxxxxxxxx               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073

---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
      joining column's datatypes do not match