April 2008 Archives
Two half-day tutorials, Help! Everyone Hates Our IT Department (And How To Change That) and Time Management for System Administrators will be taught by Tom Limoncelli at the IT360 Conference in Toronto, Canada, on April 7 (entire conference and Expo runs April 7-9). These presentations have received accolades when presented at other conferences. They've been updated and sharpened with new material. This is Tom's first appearance at the IT360 conference. Website: www.it360.ca (full workshop listing here)
In system administration we have to keep many lists: lists of users, lists of machines, lists of IP addresses, and so on. The only way to keep information from growing stale is to make sure key processes are driven off of the live database.
Here are three different techniques I've seen used:
Level 1: Periodically gather the information. A spreadsheet is great for this and simple. Once a year you collect information and then you spend 354 days with out-of-date information. I've seen this in a number of places. At Lucent they hired a company to document "everything with a power plug" once a year. The information was put into a big read-only database that everyone ignored. I wonder how much they paid for this "service".
Level 2. Automatic collection. You know that, at least for the machines you know about, data is being collected and it is, hopefully staying up to date. If the process is automated, you can run the process weekly or daily. Machines can stay hidden if the "discovery" software isn't very good, or if someone wants it to stay hidden.
Level 3. Actively-used data. Rather than storing data, if you actively use it then you know it is up to date because people are dedicated to keeping it up to date. They receive a benefit, not just you. If the inventory is used to drive software upgrades, then people will complain they are "left behind" and you'll know to add them to the inventory. If patches only go to machines in the inventory, then sysadmins are compelled to keep the list accurate so they aren't dealing with security flaws.
Level 3 is a self-correcting system, which saves times and assures far greater accuracy than other solutions.
A company with limited IP address space found itself constantly emailing its engineers begging them to return unused IP addresses that had been allocated to them. Nobody listened. When they changed the IP allocation process to be a "rental agreement" that required semiyearly confirmation that they IP address was still in active use (they received email with a link to confirm the ownership), suddenly the list became much more accurate.
The pressure for a person to keep the data should be self-serving to the person, not you. Employees are quick to demand corrections to any inaccuracies related to payroll, right? The payroll department has an incentive to pay everyone their accurate salary. The employee has an incentive to make sure they are paid the correct amount, and make sure their home address and such is accurate. I once saw a company try to send holiday cards to each employee. A secretary was about to blast email to everyone asking for their home address. Since the email wasn't going to say why she needed their address (the card was a surprise), I was sure it was going to cause nothing but a big flap about privacy. Instead I encouraged her to simply get permission to use people's home address as listed in the payroll system. While monthly paychecks were direct-deposit, bonuses and tax-info was sent by paper-mail. Everyone kept that database extremely accurate.
What do you use to keep inventories and other lists of information up to date?