in reply to Memory needed for perl process

The actual memory footprint of Perl is quite modest, compared to that of other languages (notably, PHP), because it does depend very heavily on loadable binary-library modules as well as source-code that must be used.   The Perl (5...) executable is small enough by itself to qualify as “tiny,” because very little is actually “built in.”   The actual memory footprint of a particular application can be safely judged only by inspection of a running instance.   (Could be something as simple as the Linux/Unix top command, or any process-monitor.)   It is possible that part of your colleague’s concern has to do with multi-threading, and/or the size of files that you might possibly be processing.   But, none of these purely-technical considerations will help you here.

As Corion aptly describes, you have mostly a political problem.   That is to say, a problem that must be negotiated.   Although it is possible for you to install modules locally – Google (or Super Search) almost anything about “running Perl in a shared-hosting environment” – this would not be an appropriate thing to do in this case.   (It could be seen as insubordination ... career suicide ... and given the resource constraints it is also “just the wrong engineering thing to do.”)   The Powers That Be™ have dictated the limits of the hardware that all you have to use, which strictures neither of you can control, and Informatica appears to be the primary business-purpose for this box.   Expectations as to its (Informatica ..) performance might be along some business activity’s critical path, which you probably do not know about.

You need to negotiate an implementation strategy for your project that is suitable to both of you, and to the manager(s) at all levels until, and including, the point at which both management chains-of-command intersect, since all of these managers are also legitimate potential stakeholders in the outcome, and since they will know about those critical paths in the various project plans.   The impact to production must be clearly revealed, understood and then agreed-to so that no one is ever blindsided.   I predict that the solution will largely consist of moving your project to another available box having sufficient capacity.