Re: Disk IO scheduling

From: Anthony W. Youngman <>
Date: Sat, 27 Dec 2003 22:14:05 +0000
Message-ID: <LzlBVZAtQg7$>

In article <>, hantheman <> writes
>Hi all,
>I'm looking into how databases work.
>1) Is it possible for DBMS's to schedule disk IO such that if 3
>transactions are running concurrently, it sorts the accessess by how
>near they are to eachother (to reduce the number of seeks).

Does the database control disk access, or rely on the OS?
>2) I think this is a good idea, but does this complicate transaction
>handling? What if timestamp ordering is used?

Why should it?
>3) Does asynchronous IO API's schedule disk IO this way?

Anthony W. Youngman - wol at thewolery dot demon dot co dot uk
Witches are curious by definition and inquisitive by nature. She moved in. "Let 
me through. I'm a nosey person.", she said, employing both elbows.
Maskerade : (c) 1995 Terry Pratchett
Received on Sat Dec 27 2003 - 23:14:05 CET

Original text of this message