windows allows one to partition in software, I think included in the
base OS. Can someone say if linux does or not (or at least a particular
version of linux). I do agree that hardware can do a more efficient
job than software.
I am not sure how expensive a RAID controller is to anyone who actually
needs multiple disks for performance of their app, I know my < $500
personal home machine came with it.
I believe Derby gives the OS plenty of opportunity to do parallel I/O
if there are multiple users in the database. Every thread can possibly
be doing I/O at a single time. There may be room to improve in the case
of checkpoint, but not sure how important that is.
I agree that single user Derby does not do parallel.
Øystein Grøvlen wrote:
>>>>>>"MM" == Mike Matrigali <mikem_app@sbcglobal.net> writes:
>
> MM> Yes this is probably incorrect in a number of places. There currently
> MM> is only one segment. The segment number was added in the beginning so
> MM> that if we ever had to do data partitioning ourselves we had a chance
> MM> to
>
> MM> add it without an ugly upgrade. But as you are seeing over time some
> MM> hard coding has happened and if we ever really want another segment it
> MM> will have to be cleaned up.
>
> MM> My opinion is that the OS can do better data partitioning then we can so
> MM> I don't think it is very useful to add that feature.
>
> I agree that the OS often can do a better job with data partitioning
> but that often requires expensive storage systems (e.g., RAID).
>
> I think that many applications would perform better if it was possible
> to put tables and indexes on different disks.
>
> Also, Derby does not give the OS much opportunity to optimize IO as
> long as it does not allow parallel disk accesses to a table.
>
|