Data Warehouse Migration

Business value with data warehouse migration to the cloud

Over the years, many companies have developed their own data warehouse (DW) solutions with different technologies (like Oracle or Teradata). In order to get more business value from the data and utilize modern cloud platforms, integrations and machine learning (ML) capabilities, there is often a clear preference to move away from on-premise solutions.

This is usually the case when we start a Data Warehouse Migration -project. Mostly we migrate from on-premise to cloud, but cloud to cloud migrations are done every now and then. It is important to understand that migration itself does not magically fix things like lacking processes, bad governance models or poor data models. Even though well planned cloud migration offers multiple benefits with solutions like Snowflake and dbt. 

Expected benefits from on-premise to cloud migration:

  • Good scalability as the volume of data grows
  • Improved performance with faster query performance
  • Easy implementation of Artificial Intelligence (AI) tools
  • Unlimited run-slots, environments and projects.
  • Lower total cost of operating
  • Better integrations to surrounding data ecosystem
  • Improved developer experience

Profit and benefit

 

Main thesis for data warehouse migration

Business needs to drive the transformation and capability development to create fast business value from data warehouse migration.

 

money-bag

Business needs to drive the migration

Carelessly migrating reports rarely makes sense. All dashboards and reports do not age well and they require critical examination of their necessity. Business people know the importance and prioritization criteria of migration. Low utilization indicates that the current solution is not viable for direct migration. We aim for less work and the biggest business impact possible.

map

Generated effort and impact define the priority for migration.

Data platform development is a constant battle of wise resource utilization. Available development power being limited we need to assess which reports bring the most value or are most needed. This evaluation comes from the business. It might make sense to slip in a couple of quick wins too.

screen-rotation

Launch the data application(s) into production quickly.

To validate the business benefits early on and get feedback and buy-in from the consumer you should launch the data application(s) into production early. Educate and advise users on the possibilities of end products. Observe usage and gather feedback to help assess the priorities and details of the next migration items.

Incremental migration is usually the best way forward

The incremental approach comes with several benefits:

  • Reduced risk: Completing data development and systems in smaller chunks, reduces the risk of data loss or system downtime.
  • Reduced impact on users: The impact on users is smaller since development is done in small portions.
  • Improved testing: Each new development batch can be tested before proceeding to the next, ensuring that the work is proceeding as expected and any issues are identified early on.
  • Better visibility: Better visibility into the development process, as your progress can be tracked and issues can be identified when they arise.
  • Improved flexibility: Faster adaptation to changing requirements as making adjustments to the development plan is easier.

Recordly-Pull-Ups

 

Four steps to a successful migration project

 

1. Plan the architecture, design data model and define the development practices as well as schedule 

Use senior expertise and experience from earlier projects in planning and designing the enterprise architecture as well as data warehouse capabilities. Engage IT & data people and selected stakeholders from the business units. Select technologies and cover relevant corners of migration work planning.

2. Start building selected use cases to steer the development

Build baseline DW and BI-tool for first reports by implementing most urgent and prioritised ones. Integrate needed data sources and model the DW and migrate or modernize reports. Validate together with the business consumer against the old system. Do the metrics match? What did we learn for next steps? This acts as the foundation for the future.

3. Scale the initiative

Scale to build the bulk of the work. Finish the data warehouse implementation based on the work completed in stages 1 & 2 for the rest of the business unit/stream.

4. Discontinue the old system

Test and validate the new solution and decide the final cutover from the old system. Communicate with the rest of the organization and collect further development requests.

Move to Step 2 for the next business stream.

Ask more

Mika_Pohjola_Contact.png

MIKA POHJOLA

Chief Sales Officer

+358 40 830 9486
mika.pohjola@recordlydata.com

See how Caruna made it 

Carunan toimisto 4_Lowres_NARROW

Caruna - Building a robust data platform