That's a good point. In my case, I don't think the timed read option
made that much sense because I wasn't doing any network I/O
(directly). But in a traditional network server configuration, your
suggestion makes a lot of sense. I'll have to think through the
performance implications of timed reads in this case.
===
Gregory Woodhouse
[EMAIL PROTECTED]
"The whole of science is nothing more than a refinement
of everyday thinking." -- Albert Einstein
On Jul 29, 2005, at 11:36 PM, Chris Richardson wrote:
Greg;
Actually, the HANG is a reasonable alternative, but a better
solution might
be a timed READ. The time slice can be reclaimed upon the
reception of an
I/O, otherwise, the time slice is released to others. Subsequent
global
reads are usually satisfied from Cache and don't have the big
overhead of a
physical IO. Usually only the first one is expensive, subsequent
accesses
are cheap until the value changes (a WRITE always results in a
physical IO,
eventually, but may be clustered with other changes in the vacinity
depending on the Cache flush algorythms). When you start talking
about one
node in a cluster serving as a LOCK Manager, then the rules are a bit
different. The READ must be validated by the LOCK Manager who
coordinates
the global access for the cluster.
-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
_______________________________________________
Hardhats-members mailing list
Hardhats-members@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/hardhats-members
