PMP Mail Archive: RE: PMP> Test Plan-Traffic/transaction generator

PMP Mail Archive: RE: PMP> Test Plan-Traffic/transaction generator

RE: PMP> Test Plan-Traffic/transaction generator

Bob Pentecost (bpenteco@boi.hp.com)
Fri, 24 Jan 1997 09:53:01 -0700

Chris,

I have checked with some of our test people and our tests tend to be
structured to testing one object at a time. Within the tests we do printing
and other operations, but we don't really have a programmable tool.

Sorry that I can't help on this one.

Bob Pentecost
HP

----------
From: Chris Wellens[SMTP:chrisw@iwl.com]
Sent: Monday, January 13, 1997 6:06 AM
To: pmp@pwg.org
Subject: PMP> Test Plan-Traffic/transaction generator

Some of the early plans for testing RFC1759 products included
performance tests and the use of a traffic generator.

Now that we have a better understanding of precisely what kind
of testing the IETF requires--checking of consistency and
interoperation of the values of all the MIB objects--performance
tests do not help us satisfy the IETF goal.

To check the consistency and interoperation of the MIB objects,
still requires that we have a mechanism for generating printer
specific transactions over the network in order to force
counters to count, etc. For example, we will need to cause an
alert, and then check the prtAlertIndex to see if the integer
incremented. Then we need to verify that all the printers under
test incremented in a consistent way.

Well... we have 179 objects, so this is going to take a long
time.

During the past few weeks, I have been investigating different traffic
generators (that normally generate things like "send 1500 Ethernet
packets"), to see if they were programmable to the extent that we could
pre-define a set of printer specific transactions. The idea would be to
generate these transactions, on the network for each printer, and then
check the value of the objects (probably via a tool). Unfortunately,
while these the traffic generators do permit this kind of programming,
for our application, it would be an enormously time consuming process.

So this leaves us with the alternative of doing it the old
fashioned way--having a Windows and a UNIX machine and
generating printer specific traffic through the user interface,
and we should also generate some NetWare transactions, too.

It occurred to me that some of you might already have a preprogrammed
network printing transaction generator that you use for your in-house
testing. Could we please have some discussion and suggestions
on the best way to do this?

Thanks.