ADSM-L

Re: ADSM performance

1999-09-10 08:01:17
Subject: Re: ADSM performance
From: Richard Sims <rbs AT BU DOT EDU>
Date: Fri, 10 Sep 1999 08:01:17 -0400
>Our network people tell me I should get a theoretical maximum of 4 GB
>per hour thoughput on both these connections...

Chris - If that's all your network people can offer you, then you should
        have a talk with your manager about that group providing REAL
assistance to people who come to them for assistance.  It takes no effort
for them to toss off a theoretical number.  Their *job*, on the other
hand, is real-life networking.  They should be working with you to find
out what's really happening with your network connection...whether the
device and operating system sizing parameters are optimum, etc., and
then do some real performance analysis.  (I get really annoyed when I
encounter alleged network people who can't be bothered to look for
answers.)

>and I measure at best 2 GB per hour for any one ADSM operation

You haven't told us how you've measured this or what your operating parameters
are, so there's not much we can comment on as to your realized performance.
Best measurements are taken when nothing else is going on with a transmission
line and you transmit one very large file (via FTP, ADSM, and anything else
available to you) and then compute the throughput.  Change parameters in a
controlled way and see what improvements may occur.  Your network people
should work with you to assess the measurements and recommend techniques, as
well as look at the performance of their equipment as the tests run.

>but my manager (as manager often do) won't come up with the money
>unless I can give him some specific estimates of the performance
>we'll be able to achieve with this.

This is the kind of manager exemplified in the Dilbert comic strip.
Ask your manager if he gets specific estimates of future performance from his
investment company before buying mutual funds.  We get the sense that your
manager is occupying his position principally for the income, and has no real
interest in how well the systems are serving.  It is known that boosting
network capacity improves throughput.  A manager interested in service rates
and who perceives something lacking when reviewing performance reports will
himself incite action to have the network upgraded: you as a technician
shouldn't have to be the one initiating action.

This brings us back to the performance measurements.  If the network,
operating system, and ADSM are optimally using the existing network, and if it
is not adequate to present and upcoming needs, then an upgrade is required.
Rather than specify a solution yourself or extrapolate from the technology you
are currently using, I would work with network people to determine the most
cost-effective approach, based upon all the technological choices they should
know about.  Make service problems and potential solutions known to both your
manager and those above him, giving credit to all those who helped in the
analysis.  You will at least be on record as having identified what can and
should be done, when someone higher up comes around asking why nothing has
been done about service rates.

   Richard Sims, BU   (and survivor of a large corporation)
<Prev in Thread] Current Thread [Next in Thread>