|
ntl is the UK and Ireland's largest cable operator and a leading provider of broadband and communications services. Today ntl has over 3 million telephone, TV and internet customers.
The Harmony programme began in February 2002 and now manages over 3 million accounts and bills £30 million a month.
As the data migration was taking place in parallel with the development of the Harmony systems the legacy databases were divided into logical groups and migrated, in releases, in strict order to correspond with when the new system would contain the required functionality to correctly handle the data being migrated. This added extra complication as the data was effectively being migrated to a 'moving target' which, as well as changing from week to week, would also often change dramatically from one release to the next as new functionality was added (and sometimes removed).
Further to this, most migration releases involved taking the data from several legacy systems on different platforms and moving the data simultaneously to the several applications contained in the Harmony suite, also on many platforms. The diagram below details the process for just one of the eight releases of the data migration:
During the releases Redwood Systems have been responsible for the specification, data mapping and development of extractions, transfers and transformations of the data, plus reconciliation, integrity and data cleanse activities working alongside IBM throughout.
This has involved many complications including, but not limited to:
- Duplication of unique keys across the legacy databases e.g. customers with the same account numbers but on different systems.
- Correlation of key data across Harmony applications which aren't assigned until during the load process (i.e. post-transformation) e.g. both Geneva and Harmony BAPS required their customer numbers and unique product id's correlating but the values in both cases were determined by each applications load processes and so required the development of post-load software to complete the migration.
- Databases handling data types differently e.g. in some instances the data would pass through as many as 3 different types of databases during the migration process which could cause difficulties with the way they each handled the storage of certain data types like dates and numeric values.
|
