Configuration Management for Earthlings

Configuration Management (CM) or change management is one of those mom-and-apple-pie things that no reasonable person will say much against. Of course we have to manage our resources in a documented fashion. But the way it's described in the texts and by consultants makes it seem like an unattainable dream and a maintenance nightmare all at once. They'll tell you you have to document your assets to the nth degree, and surround the entire process with elaborate "artifacts" (their fancy name for documents). Project management is similar. I once had an instructor tell us in class that good project management takes about 80% of all the effort involved in a project. While that may be good news for project management consultants, I think it's insane. No one can afford that, so no one does it, and still produce what they intended to. But while we will have projects we must manage, we can evade managing our configuration. It's ugly, but it happens. I know because I work in a data processing division with a $280M yearly budget that has not done configuration management at a high level.

Oh, we control program transmittals. But that is one of the few areas in which close tabs are kept by management of what we have out there and how it gets changed, at a level higher than the site. The technical guys are pretty good about controlling what they're in charge of and keeping it running. But our executive has no repeatable method for making decisions about what mainframes to buy or eliminate, or what sort of contract to sign this year (if any) for disk upgrades or new software. I have it on good authority that she recently lamented that she was "tired of making all these decisions by myself". Well, so are a lot of other people. She needs help. No one is knowledgeable enough make those decisions by herself (and she is definitely not an exception).

In a couple weeks, I'll know if my suggestions for jumping into CM at her level are going to be favorably received. I doubt it, but I can hope that I'm wrong--or that she is replaced soon. The charter she approved for the configuration control board had her making all the decisions, and she hasn't even implemented that. The budget is always cloaked in secrecy, and no one knows for sure what our priorities are or how much is available. There are always questionable contracts signed in late September (at the end of the fiscal year) to spend money no one else knew was there.

I'll post more on this. But for now, I have two fears: that our executive will sabotage any effort to do true CM bcause she is jealous of her unchecked power; and that the CM nazis in my agency will use the perfect to bludgeon the good into a bloody pulp, forcing us to spend all our time writing "Requirements Traceability Matrices" and the like before we can even start. You should see what they require for an "audit" or "assessment" of or CM process. Well, here's part of what they want to see:

Change Forms / Documents
Requirements Specifications
RTM
Engineering Drawings
Interface Control Documents
Schematics
CS Structure Chart
Other Technical Documentation
Deployment Management Plan
Transition to Support Plan
Computer Operations Handbook
System Test Plan
System Test Descriptions
System Test Report
Product Baseline Documentation
SW VDD and/or HW Release Record
Post Implementation Review Report

You get the idea. I'm not saying this stuff isn't all good to have. We do have some of it. I'm saying that when we're in a situation in which there is no CM process for spending hundreds of millions of dollars, it isn't helpful to demand the impossible of any attempt to get a handle on that situation. I'm for just starting, and worrying later about getting all those documents written. Even if many of them never do get produced, we'll be a lot better off than we are now.

I'll let you know how it goes.

Sorry for the Hiatus

It's mostly laziness that I haven't kept this blog up. But I've also landed in a job here that has some chance of allowing me to change the way part of IRS does business. I'll be crafting a configuration management structure that may (if the execs allow it) improve the decisionmaking around here ("here" being the organization that runs the mainframes and servers our agency uses). Hope springs eternal. I'll resist the temptation to bitch about the things I don't like about how we do things unless they shed light on government-wide problems as I see it.