Data Migration
Understanding what really lies behind the term
Data migration is a term widely used in information systems transformation projects. However, its reality is often misunderstood and sometimes confused with a simple upgrade or version update. In practice, data migration represents a much broader undertaking, involving significant technical, business, and organizational challenges, particularly in complex industrial environments.
Migrating data is not just moving it
In simple terms, data migration consists of retrieving data, or a defined set of data, from an existing information system and transferring it to another system. The source system is very often a legacy system, an older tool that is intended to be shut down or replaced.
Contrary to common belief, this is not about moving a file from point A to point B. Data is stored in databases with specific languages, structures, and rules. The target system operates differently and cannot always read this data as-is.
For this reason, a migration requires reworking the data, adapting it, and re‑establishing the appropriate standards so that it can be read and effectively used in the new information system.
The three key stages of migration: ETL
Data migration is based on three structuring stages, grouped under the acronym ETL:
• Extract
• Transform
• Load
The first stage involves extracting the data from the source system. This extraction is carried out according to decisions made upstream: whether to migrate all data or only part of the existing dataset.
The second stage is transformation. This step adapts the data to the format required by the target system. This is where mapping comes into play, a correspondence table between source data and target data.
In concrete terms, a field such as a name, first name, reference, or attribute may be named differently, limited to a certain number of characters, or structured in another way depending on the system. Mapping makes it possible to define precisely where each piece of data goes and in what format.
Once the transformation is complete, the third stage follows: loading. The data is then injected into the target system to populate the final database.
Powerful tools, but a strong dependence on data quality
To manage these different phases, specialized tools are used, particularly to automate data transformation and loading. These tools allow large volumes of data to be processed and help secure the formats expected by the target system.
However, a fundamental principle applies to every migration project: the quality of the output data depends directly on the quality of the input data. If the data is incomplete, inconsistent, or poorly structured at the start, these issues will carry through after migration.
A strongly collaborative effort
The first step in a migration project takes place on the client side. It involves assessing the existing data landscape and deciding precisely what should be migrated. Does all the data still have value? Should the entire history be migrated, or only part of it?
This phase is often an opportunity to clean up databases, remove unused data, and rethink certain practices.
Once the data has been extracted, the client must review and validate it. This includes checking information consistency, compliance with naming rules, format constraints, and field length limitations imposed by the target system.
Percall Group provides support throughout this process.
Data migration becomes a collaborative effort, closely involving both business teams and technical teams.
Sometimes very large data volumes
Migration projects can vary significantly in scale. Some involve only a few hundred or thousand objects, while others concern tens of thousands or even several million data items.
In such cases, performing a single, all‑at‑once migration is neither realistic nor desirable. The work is therefore broken down into batches, often by object type. Each batch is prepared, migrated, checked, and validated before moving on to the next.
Securing migration through testing
Before any final loading, data is integrated into a test environment, known as UAT or pre‑production. This stage allows users to directly verify their data in the target system: object searches, information consistency, and validation of business rules.
Once these checks have been completed and approved, the migration is replayed in the production environment, which will be used by teams on a daily basis.
Breaking the migration into batches makes these checks easier and allows the entire project to be progressively secured.
Expertise built over time
Data migration thus emerges as a structuring project, far beyond a simple technical operation. It directly impacts the quality of future tools and plays a key role in the success of digital transformation initiatives within industrial companies.