Re: Oracle's use of tablespaces

From: Lee Parsons <lparsons_at_exlog.com>
Date: Tue, 31 Aug 93 18:18:33 GMT
Message-ID: <1993Aug31.181833.26475_at_exlog.com>


In article <grimes.746795328_at_access> grimes_at_access.digex.net (Seth Grimes) writes:
>gerrit_at_vtm.be writes:
>
>>Futhermore a tablespace allows you to manage your database more easyly :
>>suppose you have three data disks : one for tables, one for indexes and one
>>for temporary objects. Now you can create a tablespace for each type of
>>dataobject on each disk and then you only have to specify for your indexes
>>where to go (tables will go in the default table space for the user and
>>temporary objects in the default temporary tablespace).
>
>As I mentioned in an earlier post, you can do this with Ingres without
>tying up the space -- i.e., tell Ingres to grab the space on x device *if
>it needs to* and free up the space when its done. Sorry to belabor the
>point.

What are we saying here? Is the difference that Ingress uses its own space managment system BUT has a feature that allows the db to grow and shrink dynamically. Or does it use the Native OS system.

I suspect (with never having seen ingres :-}) that they do all the managment that oracle does but allow for shrinking/growing tablespaces.

If that is true I can see how it would be useful on a system that has a bunch of different product around all requiring sort space. You could set up a ram disk that is used by all your appl as a sort area and dont have to pre-allocate a big chunk of it to oracle.

Would creation on the fly of a sort area for oracle cause a performance problem? I dont know enough about the FS code to know if creation of the sort file would be painful, but it does seem reasonable to think that precreating a sort area and keeping the free block list in the memory allows you to cut down on atleast some disk hits. I would also think that if there is anyplace in the system where shaving off a small number of disk hits is going to do you any good, it would be during the much accessed sort routine.

-- 
Regards, 

Lee E. Parsons                  		Baker Hughes Inteq, Inc
Oracle Database Administrator 			lparsons_at_exlog.com 
Received on Tue Aug 31 1993 - 20:18:33 CEST

Original text of this message