Data Migration

Data Migration Challenges

  • The target system is usually under change during the project.
  • Requirements cannot be known at start.
  • Many requirement are discovered during the process.
  • Many requirements change.
  • All result in constant and expensive change control.
  • Legacy systems contain a great deal of historical data which either has to be moved or dropped.
  • This data either has to be moved completely or migrated into data warehouses and the target system.
  • Older data requires more cleansing.
  • All this has to be performed with full auditability and reconciled.

Data Migration Outcomes

  • All data cleansed, reconfigured and moved in one operation.
  • The migration is performed in a short controllable timeframe.
  • The migration is fully audited and reconciled.
  • Edge case data is dealt with, avoiding the need for exception reports.
  • All historical data is present, avoiding the need run source systems past the end of the project.
  • The source systems can be decommissioned immediately.

Traditional Approach

  • Migration requirements are initially authored by the business team
  • Initial implementation is developed (in a traditional language/tool) by technical/business team
  • Complexity must be managed by hand in the implementation
  • All errors, whether requirements or coding errors, must be discovered and put through change control.
  • New requirements are discovered, target system changes, requirements change require costly change control
  • Reruns are only possible after the change control process has occurred and the changes have been implemented – significantly limiting their frequency.
  • Changes must be re-implemented and regression issues managed (are failures new requirements or regression issues)?
  • As the project progresses exceptions become increasing difficult to manage and usually result in exception reporting – make reconciliation difficult.
  • Final outcome is usual unsatisfactory.

acceler8’s Declarative Approach

  • Business user develop requirements stand-alone (without having to understand the implementation.)
  • Requirements are transformed automatically into the code that runs, avoid the traditional development step.
  • This delivers significant error reduction.
  • This also enables business users to easily change requirements and have the code change automatically.
  • This reduces the costs of development and change control and greatly improves maintainability.
  • Reruns can occur immediately the requirements have changed – which enables a significantly higher rerun frequency – potentially dozens a day.
  • Requirements can be discovered late into the project to deal with exceptional cases without affecting the outcome.
  • The requirements can be automatically compared to the known data models (source and target) for completeness checks and checked within themselves for consistency.
  • Reconciliation checks and regression scripts can be automated from the same sources and run after each execution to aid analysis.

How do you know … ?

  • It is important in any data migration project that you can be sure (even prove) the data was migrated correctly.
  • You need to be able to answer any question starting with “How do you know ….?”
  • The acceler8 data migration solution enables you to answer question like:
    • How do you know the requirements are correct?
    • How do you know all the data has been moved completely?
    • How do you know the data is correct?
    • How do you know the new system will operate correctly?