Perl is a general-purpose interpreted language built around text. Its regular expression engine is part of the language rather than a library, its list and hash types map directly onto the shape of operational data, and its one-liner mode makes it usable directly from a shell pipeline. That combination is why Perl became the default language for system administration, log processing and report generation, and why it still sits underneath a large amount of infrastructure that nobody has rewritten.
The language rewards knowing its model. Scalars, arrays and hashes are distinct types with sigils that change according to how you are using them; context — list versus scalar — silently changes what an expression returns; references are how anything more complicated than a flat list gets built; and special variables such as $_, @ARGV and $/ make idiomatic Perl terse in a way that reads as noise until the rules are clear. Once they are, a Perl script that parses, correlates and reports on several gigabytes of log output is twenty lines.
In a modern estate Perl appears in two roles. First, maintenance: monitoring checks, ETL jobs, provisioning scripts, build glue and vendor tooling written years ago that still runs nightly and that somebody has to be able to read. Second, deliberate use, where the task is heavy text manipulation and CPAN already has a battle-tested module for the format in question. Both roles need the same skills — regular expressions, references, file handling, modules and enough discipline with strict and warnings to keep the code maintainable.
Why this skill matters now
Perl is no longer the language teams start new platforms in, and that is precisely why the skill is scarce and valuable. The code is still running. Monitoring plugins, billing and reconciliation jobs, log pipelines, provisioning glue, telecom and financial batch processing, and the vendor tooling shipped with older infrastructure products are all Perl, and all of it still has to be understood, patched and eventually migrated.
The supply of engineers who can do that work has thinned faster than the code has been retired. When a nightly Perl job fails at 2am in an organisation where nobody can read it, the outage lasts as long as it takes to find someone who can. Teams planning migrations hit the same wall from the other direction: you cannot faithfully port a script you do not understand, and the subtle parts — context, references, special variables, regular expression behaviour — are exactly what a naive rewrite gets wrong.
There is also live demand where Perl is simply the right tool: heavy text transformation, one-off data surgery, and formats for which CPAN already has a mature parser.