Delta Migration 101

Have you ever face with delta data during migration process between upgraded systems? There are several standard approach to manage these delta. These approach can from simple to complex.

The Simple Way

The simple approach is usually relative to cost. This suits small and mid volume transactional data. You can choose to do a straight cutover from source system to target system. The focus will be on the delta data that reside between the two systems. There are two ways to handle this.

  1. Reconciliation of data between two system till the delta is cleared. These usually will take 2 to 4 weeks for the target system to reach a steady state.
  2. Stop processing of delta data in source system and to proceed in target system.

The Complex Way

If you have high transactions and a huge system. It is time to invest to manage the delta automatically. Complex way will usually involve tools or transient ETL (Extract Transform Load) integration between source and target system

  1. ETL tools are the most common approach to synchronise data at application level. It is automated and can be removed once steady state is reached.
  2. Another method is the building of a transient application to handle the delta data. This application is focus to manage the delta data and ensure a smooth transition between the two systems

Anyone and everyone will overthink Delta migration. It is always wise and wiser to consider the approach based on your transactions and budget. After all, you cannot request to fly to the moon with an aeroplane.

Agile Migration

Migration is a challenging project for many. With the Cloud model, there are plenty of migration to and fro on premises and Cloud. Some will try to migrate to a ERP type large global system only to find its constraint. Then, they will migrate it back to legacy. One common trend in these migration is often traditional approach. Instead, you can consider to adopt Agile migration.

Photo by Ethan Kwok
Traditional Migration

Traditional migration adopt incubator or sandbox approach. It involves a waterfall model of Big Bang. Initial fixes are done in the upgraded system with sample data. Changes are then deployed accordingly to testing environment for UAT (User Acceptance Testing). Once signed off, the entire fixes are deployed in a Big Bang to production environment. For each fix, regression testing are conducted again from testing environment. Duration of project will be at least 6 months to a year.

Agile Migration

Agile migration are pioneered by the Cloud trend. Lift and Shift is the most common migration approach. In this case, at least 3 cyclic runs are performed on the migrated production environment. Agile is adapted for the project approach. The duration is shorter and takes less than 6 months. Another Agile approach is incremental migration. Migration is done incremental or Agile to move from legacy to target system.

Many organisations or vendors will conform to traditional migration approach to mitigate risks. With growing trend of Cloud and Agile team, you can consider to use Agile migration. The Agile approach aims to minimise costs and duration. However, you do required experienced vendors and SME (Subject Matter Expert) to adopt this strategy.