Subject: RE: This is a software design question - ILE related

["Nathan M. Andelin" <nathanma@xxxxxxxxx>]

Chris,

Your points have been well taken.  The speed tests I did were just for
information.

Speed is not the only criteria in software design.  Availability of memory
and other shared resources, as well as complexity, potential failure points,
redundancy, scalability, and many other factors might be considered.

The question in my mind at this point is how relevant are "open data paths"?
Back in the 80's IBM warned us about them.  The manuals haven't changed
much.  They still recommend a number of ways to reduce them.  But, 12 years
ago 32 meg was considered to be a lot of memory.  In today's world of GB+
memory pools, are open data paths still relevant?

Nathan.


<snip from Chris Bipes>

I was not suggesting that the C/S was faster than read from file or lookup
in a table.  In fact I do use the table lookups for small control files such
as A/R Codes in large batch jobs where the data cannot be sequenced by A/R
code.  What I was suggesting is for the interactive program where you have
hundreds of users using the same program, to use the C/S model instead of
having the program open the file itself.  This would save the hundreds of
Open Data Paths to the file.  Yes you do have to consider the problems with
data queues when there are not enough servers to handle the transaction
volume.  This is why we delete the data queues at IPL and program the client
and servers to create the data queue at startup if it does not exist.  The
servers also allocate the queue so it cannot be deleted while they are
sitting at data queue wait.

<end snip>


+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---