System install/drive sector ?s

What you have heard is true. It is better to Install Win'95 on a newly FDISKed and formatted HD but the difference in performance shouldn't be terribly noticable.
Do you have enough (32MB) RAM in all the Machines?
Are all the CPU's, Motherboards and NICs the same?
There are countless reasons why they could be slow but usually it boils down to one of the above.

I guess the reason is the hard disks are nearly full. Copying CAB's does not change the perfrmance too much. Check hard disk space on all computers. The problem may be there.

To quote your EXACT questions:

"Is it better (performance) to have WIN 95 be the first thing that is installed on a hard drive?" "Our MIS person ... FIRST copied the Win 95 CAB files to the hard drive and then ran the Win 95 setup to install the system"

This procedure does not affect the performance of the system.  A fragmented hard drive is more noticable.

In additions to the above suggestion (which I agree with), network utilization is a factor here.  To what degree is a network being used?  Printing and workspace?  Shared applications?  Internet access?  10MB or 100MB ethernet?  Peer-peer or client-server networking? How do the systems perform WITHOUT a network connection?  

The last question should help you determine where the bottleneck you are experiencing lies.  From the info given, I would say the workstation hardware (486 generation) would be the culprit.

Ryans,

Go into the Control Panel and double click the Power icon. Turn off (uncheck) any power saving features. This may be the cause of system freezes after being on for awhile without activity.

Cumbo

Cumbo,

I think you meant to direct your last comment to mwennin....

-ryans-

Ryans,

You are right.. In too much of a hurry..

Cumbo

> I seem to recall that the reason why you want to do this is
> so that it resides on the first sectors of the drive
> which are closer to the heads and therefore accessed quicker.
> Is this true?

Hopefully, if the disk-drive hardware is functioning correctly,
the heads are _always_ the same distance away from the platters.
When a head contacts a platter, you have a "head-crash",
and physical scarring of the platter, and, of course,
the inaccessibility of the data underneath the scars.

However, note that the heads move from the edge of the platter
to the centre of the platter, in order to access each
concentric "cylinder" on the disk.

Software like Windows 98 can determine which cylinders
are accessed most-often, and can "reorganize" the files
on the disk, so that all those "frequently-used" files
are in adjacent cylinders.  This reduces the distance
that the heads need to travel, and thus reduces the elapsed-time
for reading the "frequently-used" files.

> I'm considering redoing the entire installation proceedure
> on the 24 machines and would appreciate knowledgable
> fedback (not just opinions) on this issue.

My opinion is that you should "experiment" on one machine,
and see if reinstalling makes any difference,
before you consider the other 23 machines.

Try this:
 1. determine the size of the CAB files;
 2. wipe everything from one disk;
 3. use 'FDISK' to partition the hard-drive:
    partition #1 will be a "primary" partition,
    occupying most of the available space;
    FORMAT the space on this 'C:' drive.
    partition #2 will be an "extended" partition;
    create one logical-drive within this partition.
    FORMAT the space on this 'D:' drive.
    Then, load the CAB files onto this logical drive.
    (The size of this "extended" partition depends
    on the result from step #1, above.)
 4. Install into the 'C:' partition,
    using the CAB files within the 'D:' partition.

This partitioning-scheme puts Windows at one "end"
of the hard-drive, and puts the CAB files at the *OTHER* "end".

The install may take longer, because the heads will have
to travel the largest-possible distances, going "end-to-end",
but the "installed" Windows will be entirely at one "end".

Hmmm.  Why not just take one of your current systems,
erase the CAB files, and then do a "full defragmentation"
of the hard-drive?  The result will be the same as the scheme
which I've just detailed -- Windows will be "packed" into
one "end" of the hard-drive.