Rant -- Sysadminning Is Hard

A rant -- Despite popular belief, not everyone can do my job.

(Posted 2017-04-05 15:41:11 -0400)

The science fiction writer Larry Niven used to assert that there was something about a Cadillac that caused the driver’s brain to rot. He or she came to believe that yes, the road did belong to them, and so it didn’t matter how badly they drove or how many traffic laws they broke, that was their prerogative.

I’m starting to think a similar process is at work on the brains of people who manage Cisco internet routers.

Now, a good sysadmin–Unix or Windows–will admit when they’re out of their area of expertise. Likewise for application programmers, or database administrators. But for some reason, networking types believe that once you know how to configure a Cisco device, you know it all. You can set up and maintain Unix boxes, you can write programs, you can run a database, you name it.

I don’t assert that this isn’t true. What I have trouble with are the two implicit assumptions that go along with this belief:

1) Cisco admins believe they can do all of these things well, or at least 2) No one can do them any better.

Why would anyone think this? What is it about managing a network box that makes you think you can take on everything in the entire IT world?

Today I listened while a router jockey–who is also in charge of our VPN devices and some Unix backup–explain that changing the size of a filesystem is very, very difficult. Sure it is–for you. That’s because when you set up the system, you just created one giant partition for everything, and you didn’t bother setting up logical volumes. So yeah, on your system it’s hard. And when your system’s sole file system fills up, it will crash your box, taking your mission-critical application with it.

The real solution is to design, plan, and implement a robust system.

The router-jockey solution is to set up multiple systems, so that when one fails, you can swap it out. Two (or more) systems at only twice the price (or more)!

And then when one system fails, and the one they replace it with doesn’t work because they haven’t bothered keeping it configured the same as the active system, they’ll create some half-assed, home-brewed duplication system so that the primary and secondary systems are all identical all of the time…so that when the primary’s file system fills up, it will crash all of the systems at once.

The “solution” then is to buy a bigger disk, of course.

Ever wonder why government IT spends so much money? Because hiring less experienced people is cheap in the short term but very, very expensive in the long run.

Thanks for letting me rant.

Click here to go back to the home page.