Data Center Migration Definition

Data Center Migration is the process by which we transfer data from some data storage systems to others, from one data format to another or between different computer systems.

Typically, a data center migration project carries out to replace or update servers or storage equipment to consolidate a website, to carry out the maintenance of a server, or to relocate a data center.
Depending on the type of initiative to be carried out, a different approach is necessary.

Thus, one could speak of three different ways of approaching data migration :

  1. Using a matrix-based software, which is the best option for data movement between similar systems.
  2. Relying on host-based software: that would be the most recommended option for application-specific migrations. This is the case of file copying, platform updates, or database replication.
  3. Using network devices: In this way, volumes, files or blocks of data migrate most appropriately, depending on their configuration.

These are some factors that should be considered in a data center migration project :

  • The time it will take to complete the migration.
  • Amount of downtime that will be required.
  • Business risk arising from technical compatibility problems, data corruption, application performance problems, and data loss or omission.

Challenges of Data Center Migration

Success in a Data Center Migration project will depend largely on the level of understanding that reaches about the process and its implications. Knowing the challenges involved in such an initiative is the first step. Among the most important are the following:

Storage migration:

In a project of this type, there should be no problem. As long as the application only uses general interfaces to access the data. And in most systems, this is no problem, although, in the case of old apps that run on proprietary systems, it could be. In that case, the initiative would be complicated. And also, it would be necessary to carry out the necessary tests before releasing the solution in production. Mainly for two reasons:

  • The source code of the application may not be available.
  • It could be the case that the application provider does not continue to be active in the market.

Database migration:

This is one of the simplest forms of data movement, as long as the database is used as storage. However, despite the apparent ease of the process, setbacks may arise related to:

Types of mismatched data (date, number, sub-records): In this case, we should work to maintain the integrity of the data. And it may be necessary to direct the management to the modification of some of the applications that use the database.

Different sets of characters (different encodings in each column for the same table): when this happens, you will have to review the applications that thoroughly use the database.
The ETL tools suites well for the task of migrating data from one database to another, use being further indicated in projects where there are few connections between source and destination.

Application migration:

When facing such an initiative, it is imperative to resort to a complete ETL process. This is because, even when the same provider designs the applications, they store data in significantly different formats and structures. This peculiarity complicates data transfer.

The transformation step, for example, is one of the main drawbacks. And, although relying on an ETL tool gives the advantage of its connectivity, which makes it ready to use with disparate data sources and destinations. Difficulties may occur when migrating data from mainframe systems or applications using certain forms of data storage, since :

  • Mainframe systems use record-based formats to store data that, although simple to manage, often incorporate optimizations.
  • Optimizations include storage of binary coded decimal numbers, non-standard storage of positive/negative number values or storage of mutually exclusive subscripts within a record. That is, aspects that complicate data migration.

The way to proceed would be to carry out the extraction in the source system itself. And then convert the data into a printable format that could be analyzed later using standard tools.

Character encoding:

In this case, the complexity lies in the fact that most of the systems developed in the PC-based platform use ASCII encoding. However, mainframe systems rely primarily on EBCDIC encoding that is incompatible with ASCII, and the conversion is necessary to display the data. In order to overcome this challenge, it is necessary to consider when choosing ETL tools that they must support conversions between character sets, including EBCDIC.