The first thing to do is to figure out if you even need Unicode. Converting to supporting Unicode is a very expensive and time-consuming process. Often, it's easier to rewrite the app(s) from scratch than to convert an existing app.

Second, determine which kind of Unicode you want to support. Usually, this doesn't impact very much because most (if not all) open-source solutions support all the different character sets. But, some proprietary systems may support utf8 but not ucs2, etc. There are conversions between most character sets, but it's timeconsuming to set it up.

Next is to determine if your architecture currently supports Unicode. Depending on what your systems are doing, you might have to make sure your OS(es), database(s), and other 3rd-party applications support it. If they don't, you have to start there.

After all that's done, upgrade your Perl to 5.8.4 and handle that. Before 5.8.0, Unicode is supported through the use of modules, some of which aren't the easiest to use and which provide a rather large speed hit. With 5.8.0, Unicode is natively supported in Perl. This upgrade may or may not be expensive, depending on where you are and what you use. When upgrading from 5.a to 5.b, you generally will have to reinstall every module and run your entire regression suite.

Next, you'll need to make sure that the modules you use can handle Unicode. Most can, but there are some exceptions. (I can't think of any offhand, but I remember reading about some. SuperSearch a bit.)

At this point, you're now ready to determine if your source code and proprietary protocols will handle Unicode. Good luck!