Nerds and Management November 15, 2009Posted by Chad in Advice, Tech.
Tech workers such as system admins have a hard time dealing with bosses, especially the higher up you go. We have a different way of looking at things than management does. What techies usually don’t understand is that every other group of underlings have exactly the same problems. And that management types also are just people with a different viewpoint.
A good tech worker understands this, and can see the alternate points of view. This is what makes them good. There are dozens of other tech workers lined up for your job if all you know is how to program in language X, even if you’re very good at it. This is why the guy who cross trained from a different career field into IT is typically very valuable if they’re competent.
So lets take a look at a useful scenario that a sysadmin would run across. Reporting to management the state of the network.
You have an immediate failure if you think management cares about what you do. They don’t and shouldn’t, since that’s your job. Since you’re reporting up the chain, it’s your job to care about what they need to know. That’s money and problems. Why should they care about the minutia of memory and RAM. When preparing for your report, here’s a few items that they do care about and how to do it.
- Capacity planning: “Based on our performance monitoring over the last 6 months, (show the charts you’ve made) we have 4 systems that will require additional disk within six months at a cost of $x, and two that need additional memory immediately. My projections go out to a year based on current growth rates of our service.”
What, you don’t have performance monitoring statistics? You’d better start now. And learn how to do projections based on business growth rates.
- Bottlenecks: “When we look at the end to end system, it seems that our primary bottleneck is in our backed SQL database. Transactions take 1.2 seconds to complete, while the rest of the system completes in .2 seconds. We’d like to spend $$ on a DBA to go over and optimize the queries.”
Learn to use performance monitoring tools if you haven’t already More importantly, learn what they mean. Find the bottlenecks and work with developers to fix their problems. You probably won’t be able to tell them that one sorting function is faster than another, but you can say that the architect who thinks running 30 web front ends against an old single celeron server with a gig of RAM running the back end database is probably a bad idea.
- Costs: “Running our entire data center costs $y per month. The electric company is due to raise rates in 3 months (yes, you call them and ask!) so our costs will increase to $z. We think we can offset that cost by calling the server manufacturers and ensuring that our warranty will cover things if we reduce our air conditioning by 3 degrees. We also have 6 servers coming off of warranty in 6 months. The cost to replace them will be $x which includes migration costs, and the cost to extend the warranty is $y. Or we can start looking at virtual servers. Most of our systems run at a constant 3% CPU and 1/3rd the available memory. By consolidating, we’d get rid of half of our oldest and least performing servers, saving $z…”