Re: Locally managed tablespaces and migrated rows

Actually, chaining and migration are not quite the same thing. For purposes of this discussion, however, the differences are meaningless.

LMT with uniform extents are *not* disasterous. Space management within a segment has not changed as of 8i, so no fears there. The tablespace simply manages extents without use of the UET$ and FET$ dictionary tables. If you want to avoid serious ST enqueue contention, use LMT and quit worrying.

Uniform extents are available with or without the bitmap tablespace feature. In my experience, extent uniformity is a good thing. In a large scale environment you can set up a number of tablespaces with standardized extent sizes and place objects in them accordingly. When one object either gets dropped or truncated, all other objects within that tablespace are able to make use of those freed extents. Forget about tablespace reorganizations for all time.

On the other hand, in a small scale environment... well, who cares?

As one wise dude recently said, "It's the future. Get used to it".

• ricky

Nuno Souto wrote:
>
> Or chaining, as we used to call it. Probably already answered, but I
> can't find anything in google.
>
> Locally managed tablespaces with uniform extent allocation sound to me
> as a recipe for disaster when tables get many migrated rows.
>
>
Received on Mon May 07 2001 - 20:01:24 CDT