12-07-2012, 01:37 PM
DATA MIGRATION
DATA MIGRATION.docx (Size: 2.32 MB / Downloads: 57)
PLAN OF WORK
How do we migrate data?
Once the decision is made to perform data migration and before migration can begin the following analyses must be performed:
• Analyze and define source structure (structure of data in the legacy system)
• Analyze and define target structure (structure of data in the new system)
• Perform field mapping (mapping between the source and target structure with data cleansing, if necessary)
• Define the migration process (automated vs. manual)
Column Profiling.
Column profiling analyzes the values in each column or field of source data, inferring detailed characteristics for each column, including data type and size, range of values, frequency and distribution of values, cardinality and null and uniqueness characteristics. This step allows analysts to detect and analyze data content quality problems and evaluate discrepancies between the inferred, true meta data and the documented meta data.
Dependency Profiling. Dependency profiling analyzes data across rows ¬ comparing values in every column with values in every other column ¬ and infers all dependency relationships that exist between attributes within each table. This process cannot be accomplished manually. Dependency profiling identifies primary keys and whether or not expected dependencies (e.g., those imposed by a new application) are supported by the data. It also identifies "gray-area dependencies" ¬ those that are true most of the time, but not all of the time, and are usually an indication of a data quality problem.
Redundancy Profiling.
Redundancy profiling compares data between tables of the same or different data sources, determining which columns contain overlapping or identical sets of values. It looks for repeating patterns among an organization's "islands of information" ¬ billing systems, sales force automation systems, post-sales support systems, etc. Redundancy profiling identifies attributes containing the same information but with different names (synonyms) and attributes that have the same name but different business meaning (homonyms). It also helps determine which columns are redundant and can be eliminated and which are necessary to connect information between tables. Redundancy profiling eliminates processing overhead and reduces the probability of error in the target database. As with dependency profiling, this process cannot be accomplished manually.
FUNCTIONAL REQUIREMENT
Hardware Specification:
• 400Mz Processor
• 512MB RAM
• Pentium 4 CPU
• 80 GB HDD
• Corporate LAN and Internet