Path: utzoo!mnetor!uunet!husc6!cmcl2!rutgers!columbia!garfield.columbia.edu!andy
From: andy@garfield.columbia.edu (Andy Lowry)
Newsgroups: comp.databases
Subject: Concurrency control info wanted
Message-ID: <5535@columbia.edu>
Date: 19 Apr 88 16:30:03 GMT
Sender: nobody@columbia.edu
Reply-To: andy@garfield.columbia.edu (Andy Lowry)
Organization: Columbia University CS Department
Lines: 35

I am looking for information and references on the following topics:

	1. Effect of concurrency control on overall performance of
	database systems.

	2. Implementation of concurrency control in ACTUAL database
	systems.

With regard to #1, I would also welcome more generally any information
on the breakdown of processing costs among the various tasks of DBMS
(besides concurrency control, this might include logging and other
tasks in support of recovery, query parsing and optimization,
relational operations (join, projection, etc), index lookup and
maintenance, etc.).  With regard to #2, I'd like to find out which
policies (two-phase locking, timestamp ordering, conflict graph
testing, ...) are actually used, how they are implemented, what
factors affected the choices, hindsight, etc.

For both questions I am interested in centralized and distributed
databases, and also in database machines that support concurrent
queries.

I have many references on proposed concurrency control policies and
algorithms.  What I'm interested in here is: what do things really
look like in the real world?  Thus commercial systems are of interest,
but also working non-commercial systems.

Literature references and personal knowledge are both solicited.
Please e-mail directly to me, and I will summarize to the net.

Thanks!!

-Andy

andy@cs.columbia.edu