Measuring the Quality of Work of a System Administrator [closed]

Solution 1:

You should only ever provide a formal rewards system if you want your technical staff behaving like lab monkeys, pressing buttons to get their treats.

The only real metric is how satisfied you are with your admin's performance. Any metric will optimize your admin's behavior for the metric, regardless of how dumb or short-sighted the metric is. By establishing a metric, you will invariably run into the situation where doing the right thing involves ignoring the metric, because the metric doesn't take into account the precise situation or consequence.

At which point, you're only hope is that you've either got Dr. House, Network Engineer, or your newest and least experienced person in charge---either the meta-genius who is willing to throw out your process when the situation demands, or the n00b who hasn't been there long enough for your process to sink in.

The likelyhood of you having a House and a strict process at the same time is pretty slim (they are all working for startups where the concept of a performance review is alien to them anyways), which leaves your only hope being the miracle of the guy in charge being totally new to your company, but familiar enough with your systems to resolve the issue at hand.

Solution 2:

It's tough to come up with a generalized measure since it's rare that any two sysadmins at different companies do the same thing. One can, however, come up with a personalized set of measurements by defining the specific responsibilities of the position and determining metrics that will speak to them.

• If the sysadmin is responsible for production systems, uptime of the system can be used, the number of preventable incidents, or even the response time (and/or resolution time) to those incidens.

• If the sysadmin is customer-facing, one could use customer satisfaction, number of tickets closed or some such.

Bottom line, pick what you want to encourage and determine metrics to guide that behavior.

Solution 3:

This is not really an answer to you're question, but I feel I must butt in and say it since I get the vague feeling you're trying to put a number beside how good a sys admin is.

Measurements - This one by Joel is a good one

Sins of Commission - Very good article by Joel... read it!!!

Essentially it comes down to, when you try to measure something like how good someone is at their job, whether you provide an incentive, or just want to know how good they are, you always loose in what you're not measuring.

I'll provide an example, at work (large Canadian telecom) we have an outage database and track weighted outages throughout the year, and have an objective in our "performance reviews". We also have an outage notification system where internal people can see essentially what's broken right now. Now because this is tied to performance and bonuses, I've seen people outright lie on the outages so that it doesn't weigh as heavily against them.

Now I discovered our brand new untested firewall was dropping packets, causing tunnels to drop intermittently, and some other minor stuff. I opened the outage ticket as low or no impact, but said customers could see intermittent drops or throughput issues, but it would be rare.

A week later and someone doing performance testing from another group and I were chatting, and he was telling me about sometimes we're getting bad throughput compared to the other carriers, and I simply said well that's likely due to the packet drop issue we have. He had no idea it was going on, and sure enough when I checked the ticket my supervisor had closed it the next day as not an issue. We went around him and notified upper management, but it makes me wonder how often this has happened. I know employee's who literally won't touch stuff because they're afraid of being chewed out by their manager the next day if they break something.

Once a year or a statistic is not the way to monitor you're employee's. Up time may need to be measured due to SLA's, but honesty trumps all and you need to reward good behaviour, and the only way to do that is to do you're own job well, work with you're employee's on root cause analysis and outage notifications, comment immediatly on positive or negative feedback you have, don't leave anything waiting for the bi-annual review.

More to what you may be looking for is to talk to those people who you're admin interacts with and see if they respond kindly and curteously, or if you're the group everyone hates dealing with. If there's an outage, instead of measuring the uptime, get the sysadmin to explain what happened and why, and what can be done to ensure it never happens again, and follow up to see that the fix is done.

Solution 4:

Does your sysadmin:

• Keep the systems running?
• Prioritize well?
• Keep you informed of what you need to know?
• Complete projects in a timely manner?
• Provide accurate estimates of time and resources needed to complete projects?
• Explain things in an understandable and appropriate way? (Without condescension?)
• Maintain confidentiality and an at-all-times ethical demeanor? (In other words: do you trust your sysadmin?)
• Document systems well enough that a qualified person could operate them in his stead?
• Continue learning as the technologies evolve?
• Inspire confidence, even when the wolves are at the door?

If yes to all, then your sysadmin deserves a passing grade at minimum.