Phases of Data migration: Test

Before any data is moved, it is important that some portion of the migration plan be tested and validated. Results of the migration test determine whether modification of the migration plan—for example, time line, migration tools used, amount of data migrated per session, and so on—is required. For example, if testing shows that allowable downtime would probably be exceeded, the migration methodology needs to be revisited.

Testing is a key element of the overall lifecycle of the project. Why? It

(1) proves the capability of migrating the data with no impact to the enterprise

(2) provides a good understanding of risks

(3) provides the ability to accurately define the sequence for the final migration including timelines

Start this process by reviewing the outputs from the analysis and planning stages. Engage a cross functional team and assess the capabilities, knowledge and expertise of the team. If the team has the skills, knowledge and expertise and has gone through a similar exercise in the past, then this phase is greatly simplified. Follow the same sequence as identified early.

In most cases, you might have to start from scratch. In that case, start by outlining the dependencies for data extraction, sequence them in the proper order, once the data is extracted from the source system, validate against the target system to ensure data integrity and then proceed with a sample load.

This is the time to engage your IT administrators to the fullest. Review application, database, network and infrastructure architecture and optimize from a data migration perspective.

For e.g. most data migration projects involve persistence in databases but this activity needs to be kicked off from the application layer following a syntax and methodology involving some structure in flat files (txt, xml etc.). In this case, the application and databases need to be tuned to identify the right parameters which will enable you to accomplish the load in a timely manner.

If you have distributed or federated systems, you will need the assistance of network and infrastructure administrators/architects to tune the network and servers from optimum performance, for e.g. remove bottleneck processes or establish a dedicated network etc.

This phase doesn’t conclude with successful migration and establishing a proper timeline for go-live. It should also include testing of post go-live activities. In most cases, search engines will need to be updated so that the indices are refreshed with the newly loaded data. there a number of such related activities that are tied to post go-live which are usually overlooked causing performance nightmares upon start up after data migration.

Keep at it, you can almost see the light at the end of the tunnel, next phase is validation. Remember the mantra “I love data”

"Disclaimer: The views and opinions expressed here are my own only and in no way represent the views, positions or opinions - expressed or implied - of my employer (present and past) "
"Please post your comments - Swati Ranganathan"


Post a Comment