Started as a sysadmin 12 years ago. Money is great, better than a corporate programming job, and if you look at the role as a automation programmer role for infrastructure software you'll have a better time of it. Don't just fix it, make sure it never happens preemptively/again then prove that to whoever's paying your bonus.
The downside is working with other people's shitty software, particular in the enterprise space where ticking boxes is more important than security or other features. There are 'SOX compliant' banking environments where the unchanged root password is stored in plain text on a file accessable by any client as a Windows drive. But they have a security policy exemption so it's OK.
Personally I can't stand the legacy stuff - working on something that's being retired gives me the shits and the ancient userspace of the proprietary Unix sucks when compared to the modern ones - but if you want to learn Solaris (especially Solaris 8, the most widely deployed release - yes, there are more people talking about ZFS and dtrace than using either), you can make some extra pain money for dealing with it till it's gone.
Re: proving your worth, I started as a knowledgable nerd helping the senior sysadmins, at somewhere where there was good comraderie - some of the people I met there became friends for life and by helping others rise up later in my career I did the same.
Major growth areas: virtualization (you can make great bank very quickly in vmware, then transfer that into cloud skills learning xen or kvm & cloud deploy concepts in your own time), storage (but it's boring). Anyone with web reporting skills will will points with management, as will anyone who can produce and consume microsoft file formats (eg, reporting, someone sends me a spreadsheet with 390 static routes, you turn that into a routes file). Your manager wants something they can show their manager to prove they (ie you) are doing a good job.