Revenge of the Developer

Ah, the developer from yesterday was back today. It started with a harmless IM message: “Are you busy? Could you come over to my office to talk to my manager and I?”

Today’s gonna suck, might as well make it bad right away.

“Well, we’ve decided we can make this application cluster-able. We’ll use an Oracle database hosted by the DBA group instead of a local MySQL database.”

“Oh, I didn’t know you needed a database. Okay, Oracle is fine — we preload the clients on our Linux hosts anyhow. Are they going to cluster the database for you?”

“Cluster the database?”

“Yeah. Make it highly-available. You told me that this application needed zero downtime. Databases need patching, too. In fact, Oracle’s had some serious holes in the past couple of months where the databases needed to come down and get patched.”

At this point the project manager steps in.

“I don’t think we’re asking for zero downtime, just reasonable assurances against outages.”

“Those assurances come with price tags. We can do a lot to prevent outages, but we also need to account for maintenance and patching and reboots and stuff.”

“Reboots? For what?”

“Operating systems need to have security and bug fix patches applied regularly, or they stop working.”

“Well, yes, but those will have to happen when we update the application every six months, as per our schedule. The customer agreed to that.”

“It would have been better if you’d checked with us before you promised things to customers. I’m pretty sure we’re not going to be able to deliver what you promised them. Security patches don’t wait six months for a maintenance window.”

“Why not?”

“Because you get hacked right away.”

“Why?”

“Because hackers learn about the holes right away, and use them right away to break in before you patch.”

“We haven’t had anything that’s been hacked.”

“Yeah, because we’re super vigilant about patching! And we have had stuff get hacked. And we can be vigilant because everybody else let us build their systems to accomodate reboots and a little downtime.”

“We want to limit downtime — our customer wants this application to be highly available.”

“Yes, I had heard that. Say, look at the time. Why don’t you send me an email with the specific requirements of your application, and we can work from there?”

*sigh* Let’s see what they send.

Note to self: mainframe programmers don’t get it.

Second note to self: get the desktop support guys to repossess this developer’s Linux box. They owe me a favor for figuring out who was crashing their printers.