If the data is as trivial as it seems, split /:/, ... will probably be fine. It seems that's what you've decided to go with.

If the data format spec allows for things like quoted fields, and embedded colons or newlines within quoted fields, then a proper CSV parser would be preferable. Text::CSV or Text::CSV_XS can be configured to use a colon as the delimiter, and to permit quoted fields, escaped delimiters, and embedded newlines.

It's probably irrelevant; your data may never become "tricky." If it does, be aware of the potential pitfalls, and of the options available to you. It might be wise to design your application's input parsing with a very thin abstraction layer so that it would be easy to plug in a real CSV parser if it becomes necessary in the future.