On the contrary, I'm unable to to think why relational database server (or embedded solution) would be
a) more reliable
b) faster
c) easier to maintain
d) faster to develop
e) cheaper
f) use less resources (RAM)
than hard-coded perl module. This data is static, and seldom changes (once a year?).
The thing is that we do need network based (e.g. HTTP) lookups for other server to make, but even more important is some kind of library (a module that is), that will be used by other (perl) scripts (web and backend) for batch processing.
E.g. a CSV file with 50k phone numbers to be imported over the Net, and processed. IT doesn't make sense to do SQL query for each single number. A library module is needed.
Also, we simply do not use relational DB server for this project, and it's not worth to set up one, just for these lookups.