system admin staffing requir.(35unix ser

Funny, I write job requirements for Unix admins at least once a week. Never thought I'd see a question on it. 1 sysadmin could easily handle 35 servers. I am managing two sites now for my company (one with 55 and the other with 40 - 50. The average unix admin is reported to manage 60 - 80 servers. So many things can be automated that it isn't very time consuming to do day to day maintenance. I wouldn't focus so much on the IBMs in your search. Unix is unix for the most part. A good Solaris admin will pick up AIX fairly quickly. I learned AIX on the fly in about two weeks after doing Solaris for five years. Most of the applications you list are also heavily used in the Sun arena. I would suggest searching for someone with 3 to 5 years of Sun experience (mid level) or 3 years of sun specific experience and an overall of 5 years in Unix. dealing with the range of servers you mention (e3000 to e6500). I would also list your Unix specific applications, like you did here and require that the person know about 3/4 of them. I suggest starting with this (just one) admin with the range of experience I listed. With only 35 servers (22 critical) one admin should be able to handle the load and the on call duties. He/she shouldhave these servers humming and not need to be involved on on-call status too often. I manage about 100 (all mission critical) servers now and get paged about once a month. However, not all environments are the same and the amount of on call duty will depend on the way applications are managed and the amount of transactions. Like I said, I suggest starting with the one, but if the on call load is too much you may consider bringing on a junior level (1 to 3 years Unix experience, not importantwhat applications he/she knows) admin that your mid level admin could train and the junior could do day to day, non-project stuff and share the on call load.
I hope this helps.

So much depends on your definition of 24x7 coverage. You also need to look at response time for your mission critical functions. Vacations and projects also come into play.

If you need immediate response you will need someone onsite 24x7. That's 4 people if they are flexible about scheduling.

If you can stand an hour or two between the failure and the first response, you can get by with two people. That allows one to be on vacation, and also allows for continuity of business knowledge when one burns out and changes careers. It also allows one to handle day to day problems while the other is handling projects.

I like three as the minimum number for a good sized shop. With three you can afford to train people, they don't burn out as quick, and things (user support, projects, monitoring, etc) don't grind to a stop each time there is a hardware failure. Each person is only on call every third week, so you get better responses to after hours problems and less burn-outs.
I, for one, will never again take a salaried position as the only system admin.