UNIX for Dummies Questions & Answers

I was asked to maintain more than 160 UNIX servers which OS are Solaris, linux,susue,,etc by the president. But there is only one UNIX engineer in my section. I am sure I need at least 3 UNIX engineers to maintain more than 160 UNIX servers. What do you think how many engineers I need?

It depends on a lot of factors. At least two for sure. What happens if you break a leg or something? A 1 to 100 ratio seems reasonable. We are at 1 to 135 or so... we could really use another person. We have stuff pretty much under control though. If you have some problems, maybe you need 3. 4 starts to sound excessive.

One bad scenario is a power loss to the computer room. (Despite a super ups backed up by a diesel generator, this happened to us!) How long would it take you to restart everything? Can your company live with that?

I would be interested in other peoples' opinions.

Break a leg? How about just taking a vacation?
There are ratios that you can find on the Internet (such as this article ) which may help in getting a few more folks. Realize it hardly ever comes down to 'the perfect mix'. New projects come up which add more servers than you can handle, folks leave, company downsize both personnel and systems. It all comes down to what the company can afford (or what they can get away with).

I am working on the same computers since 1999 - there were approximately 150 servers when I started and 16 people (we also did user support). Then cut-backs - 150 systems, 8 people (still doing user support). Then user support came down to almost nothing. Then the company was bought out - we ended up with 4 people, 150 servers. Then we were down to 2 people. That's when they started getting rid of servers - down to 40. Also, during all this time we had 24 X 7 coverage on-call. Then the servers and SA's were 'sold' to another company - now we are at 10 sysadmins and 400+ servers (with around 25 of the original servers from 1999).

End result in my opinion - push to have at least 3 people - especially if you have to have 24X7 support.

I think the degree of automatic supervision of platforms and processes is a key factor. I believe some people call this system management. If you have good and reliable automated system management practices in place, including automated backup and recovery, process and file system supervision, network and host based IDS, managing a large network can be easier. I recall, years ago, managing many hundreds of devices, including UNIX-based services, routers, etc., across a tiered approach of operators, first level support and secondary expert support for problems that cannot be solved by the rest of the team.

Automatic doesn't always help - those original 150 servers I wrote about - many of the updates were done automatically. Until a admin made one typing error and instead of updating /etc/passwd with a new file, it overwrote /etc. 46 servers down in a matter of minutes. To err is human - to really screw up you need a computer.

We have one security admin, a network admin, 3 HPUX admins, one linux admin. All of our boxes are 24x7. We have about 25 HPUX midrange boxes, an itanium box, and 123 linux boxes.

Because of the way things are divded up, everybody works on every box now and then. But specialization does take a toll - the security guy was not allowed to take off on vacation for 2 years. He quit a month ago. The other folks are taking up the slack right now.

wow!! that was so informative. I would like to see more comments on this thread. It is a real interesting one.
Can there be some standardization. I agree that it is going to be company dependant, work dependant and offcourse cost dependant. But if this discussion comes up with a solid generalisation, a standard or a thumb rule ( as perderbo said 1 engineer for 100 servers ) it will be great. This concept could also be further extended not only to unix engineers, but probably projects too

Lets see if we get more interesting posts on this thread.