Manually Creating Shared Tables and Data Files

There are lots of ways to import Shared Tables from other tools and other kinds of metadata — but sometimes it’s helpful to just “do it yourself.” You may havehome-grown spreadsheets or documents that contain your column metadata, or simply don’t have immediate access to the COBOL FDs, XSDs, ERwin (or other popular modeling tools) models, or other standard forms of metadata that contain critical table and column information.

See this link for more information on Shared Tables and Data Files… What are Shared Table Definitions?

A new utility (courtesy of our creative engineers) has been built that supports the manual creation of Shared Tables and Data Files within the repository. This provides new flexibility for dynamically, and programmatically creating these structures for data lineage and impact analysis purposes. The utilty takes advantage of the istool archive format and syntax. Attached to this posting are three files…the documentation, a .jar file to be copied to your istool directory structure, and a sample import .csv with a variety of Shared Tables and Data Files. It will work with istool in both the Server and Client side environments. Please see the documentation for further details…

Notes!

1) After download, remove the .jpg suffix for the .jar and .csv files !!!! (the suffix is there just to satisfy this blog’s restrictions)
2) I noticed that the .jar file had its dots (.) changed to dashes (-). Leave the underscore, but rename the file so that each dash becomes a dot. The actual jar file you copy into your subdirectory should be: com.ibm.istools.cli.generate_1.0.0.jar
3) USE/READ THE DOCUMENT. It illustrates the detailed syntax that you need.
4) I noticed on linux that a ‘/’ was appended to the .isx file. I renamed the file before doing the “import” step. It worked fine. It seems to work fine “as is” on Windows.

Documentation
jar file [remove ".jpg" after download]
Sample .csv file with detailed table metadata [remove ".jpg" after download]

(what are?) Shared Table Definitions

Information Server 8.x introduced a concept called a “Shared Table Definition.” This is a “container” for Table and Column metadata. It holds the Column names and other critical metadata (sql data type, length, etc.) for your relational tables. We call it a “Shared Table” because it is automatically shared among all of the tooling within Information Server and Foundation Tools.

A Shared Table is immediately visable (and usable) in Information Analyzer, FastTrack, Metadata Workbench and Business Glossary. It is also available for DataStage and QualityStage via the Metadata Sharing option within the DataStage Designer. Shared Tables are “like” DataStage Table Definitions, but with a broader, more enterprise scope. DataStage Table Definitions are still supported as always, but are unique to the DataStage environment. Inside of DataStage, a Shared Table appears in the Repository listing with a little white “book” icon above it. I always call it the “dove” icon because it looks to me like a bird.

Shared Tables often reflect a “real”, physical table. They have a multipart name that includes Host, Database, Schema and Table and finally Column names. Shared Tables can be created in a variety of ways:

– Importing tabular metadata via “Connector” inside of FastTrack, Information Analyzer, or the DataStage Connector Wizard (under import…table defs).
– Importing tabular metadata via the Import/Export Manager using Bridges and/or Brokers
– via the “Create Shared Table from Table Definition” option from within DataStage
– via a new utility that allows the creation of Shared Tables and Data Files (also known as Shared Files) from .csv [saving that subject for another blog entry]
– via export of physical models from InfoSphere Data Architect (which could have imported the model(s) from other design tools)
– via Optim export from InfoSphere Discovery

Once they are created, you can see them in the “tree” views of FastTrack, IA, and Workbench, and also delete them in most of those Information Server clients…..each client is looking at the exact same copy of the metadata — so there is absolute consistency between the views, regardless of which client you are using or which role you play in the organization.

These Shared Tables drive the integration activity, or the review of such activity, throughout Information Server, and they also serve as an important “anchor” for concepts that help enable strong information governance. Shared Tables retain associated profiling information (domain population details, data quality assessment history, etc.) from InfoSphere Discovery and Information Analyzer. Shared Tables can be assigned a Data Steward — a subject matter expert who is most familiar with a particular table and its value to the organization. Shared Tables and their Columns can be assigned to a Business Term — or to multiple Business Terms….to futher define them and qualify their purpose and meaning to the enterprise. Perhaps most important, Shared Tables are the key sources and targets for Data Lineage reporting and analysis. Shared Tables are “connected” for data lineage purposes via DataStage Jobs or Extension Mappings that represent processes external to Information Server and Foundation Tools.

Related to a Shared Table is a “Data File”. A Data File is also a container for Fields (or Columns), but is associated with a sequential or flat file instead of an RDBMS. A Data File has similar characteristics, and participates in data lineage and metadata enhancement (Stewards, Terms, etc.) like Shared Tables, but has a different icon when viewed within the Metadata Workbench. The naming convention for a Data File is Host, DataFile name, and Data File Structure….and then Fields. Today, Data Files are uniquely used by DataStage, and created from there using the “Create Shared Table from Table Definition” as noted above. The difference is in the “Locator” tab of the original DataStage Table Definition. If it says “sequential”, then a Data File will be created instead of a Shared Table. However, a new utility just became available that allows Data Files to be created at the command line. I will be writing another entry in the blog to describe this new utility that provides new flexibility for creating these useful objects.

Ernie