|
The process of moving data from its current location, typically requiring transformation to allow it to be used in the target system. This is a very common requirement during Information Architecture projects, for example defining Classification schemes often requires that an existing list is adjusted so it can be utilised. FrequencyThe majority of data migrations are "one-off". That is the data is transformed and used. This allows the implementor to take a number of short-cuts that would not be acceptable in a more systematic migration. Even in this situation, however, it is often valuable to be able to relate the final values to the initial ones. This is because of the two different sets of corrections that may arise in future, corrections to the migrated data help to suggest ways that the source can be upgraded and subsequent corrections to the source data may help improve the migrated information. Keeping track of "audit" values so the two data sources can be related to each other will often make this process simpler. The other extreme occurs when two sets of data are to be synchronised. For example if a PDA and laptop both contain lists of contacts, and both potentially can read and correct the entries. Implementing such systems can be complex and time consuming. There are many situations which fall between these two extremes. The crucial element is to be clear about which repositories can be updated and by whom. Once this has been clearly defined the data flows necessary to enable the system to work can be quickly identified. TransformationsAll data is held in its own data form. Any data migration must transform the data at six different levels:
To pick an example let us suppose we have a set of weekly music charts on a web site and we need a list of the artists mentioned and a count of the number of weeks that each had in the charts. The source data is accessed via HTTP as a set of pages one for each week. We need to convert it so that it can be inserted into a relational data model. So we have to convert at each of the six levels, it is usual to start at the lowest level one:
These transformations are complicated enough even if we are dealing with a "one-off" migration. If there are elements of synchronisation they have to be performed in a systematic way. ToolsThere are a wide variety of tools that can help with data migrations, for example:
Links to this pageThere are no links to this page |