A detective with a magnifying glass

Validation best practice is an important topic that is often not given the importance it is due until it is too late. It is skipped by the undisciplined, hurried over by the impatient, and pushed aside by short-term-cost cutters. And sometimes, it just happens, like when experience is lost due to the retirement or turnover of skilled staff.

The basic question is (or should be): how do you know your system works as designed, and, in a disaster event, recovers as designed? Here are some ideas and tips on a few types of validation:

  • Deployment Validation – the most important validation. You need to know your production deployment works as designed.
    • Take time to write down a validation plan which includes sections for entering validation data.
    • Decide in advance of the deployment what action to take when validations fail.
    • Record the validation output and/or screens.
    • There is no short-cut to discipline. Stick to the plan, no matter how tempting it is to cut corners.
    • Schedule significant time and resources for this step.
  • Directory Synchronization Validation – one of the trickiest, but equally important. Recon the recon?! Exactly!
    • All directory synchronization tools are not perfect and highly subject to exceptionality, which is why most directory synchronization products run “full” reconciliations. But often the full reconciliations also are not perfect. How do you check this and find the out-of-sync records? You should consider a synchronization verification tool; essentially a second reconciliation tool, but read-only and specifically designed to double-check the primary reconciliation tool.
    • A second reconciliation tool sounds impractical, but in practice, is almost always necessary.
    • Go lo-frill and maximize on precision and simplicity.
    • The more customized the synchronization, the more need for a second reconciliation tool.
  • Data Restoration Validation and Drilling – a basic, often overlooked validation. Can you restore your data from a given point of time if you lose it?
    • Test restoration of a randomly selected set of data from a given point of time.
    • Validate the integrity of restored data with whatever application that consumes it.
    • Drill it! Test the procedures at regular intervals. Regularity brings familiarity. An emergency is not the right time to become familiar with procedures.
  • Disaster Recovery Validation and Drilling – don’t let it be too low of a priority until it’s too late.
    • It’s always easiest to validate your DR before you go live. Once in production, DR validation becomes a lot more difficult and risky.
    • Costs and time to delivery increase with DR validation, but the investment is worth it.
    • Drill this one too. A disaster is no time to try the procedure out the first time.

© Copyright 2020 Rex Consulting, Inc. – All rights reserved