Thursday, June 11, 2009

Teiid 6.1.0 Released, adds Hibernate Dialect for Teiid

We are excited to announce the availability of Teiid 6.1.0 along with a coordinated release of Teiid Designer 6.1.0.

There are numerous fixes and improvements in this release. See the complete list here.

Some of the highlights of the release include:
  • Simplified embedded deployment and configuration. Deployment of embedded in JBoss AS 5.0, Eclipse DTP, Squirrel, etc. is straight-forward. See some examples on our wiki.
  • Resolved class loading issues surrounding the usage of vendor specific drivers.
  • A "Hibernate Dialect" for Teiid is available. Accessing the your "virtualized" data is now as easy as creating a VDB and writing a Hibernate application.
  • Enhanced processing with the introduction of a partitioned merge join, sort optimizations, and a variety of new push down features.
  • Stream based processing for XML connector.
Enjoy the features, please be sure let us know what you think. We always welcome your comments and suggestions!

Thank you.

Teiid Team.

Thursday, June 4, 2009

A Relational Data Integration Engine

Teiid has been an open source project for few months now and the response from community has been fairly positive. New users are taking it for test drive and we have busily been executing on our roadmap. Still I sometimes feel, we may have not explained what Teiid software really does?

Since I am java six pack, I am going to dumb it down to level that I understand, and present it to you.

If you look around Teiid site depending, you will find a multitude of terms and definitions, here are some examples

  • It is a EII (Enterprise Information Integration) tool

  • It is a Virtual Database

  • It is Data Integration Engine

  • Data Mash up Engine

  • Data Virtualization Engine

All the above are accurate. But, I want to describe Teiid's base processor more precisely.

"Teiid is a Relational Data Integration Engine"

Using Teiid and its supplied tools, you build a brand new logical data model for your data in multiple heterogeneous data sources using SQL as glue language. Now, the schema from this logical model is what you can expose to your end applications as domain model.

So, in effect you took multiple existing data sources, and defined relationships among the data you have, to create a data model you need without much effort, i.e. you got model driven data integration.

Once this logical schema is deployed into a Teiid runtime engine, Teiid behaves like a relational database. It provides a secure, preferment and transactionally aware system that can be accessed by your end user applications. Note, that the physical location of your data still be in the same place as before, however the logical model results are built at runtime for you by Teiid.

Well, all is good so far but I said, "Teiid is relational data integration engine" does this mean Teiid can only integrate data from multiple relational databases? absolutely not! Teiid can integrate data from any RDBMS like Oracle, SQL Server, DB2 etc. as well from Text files, Web Services, Excel sheets, XML, Main Frame or any of your custom data. If you are wondering how that is possible? The answer lies in Teiid Connectors.

Teiid Connector: A connector is data access mechanism to a "source" system like DBMS, file or a web service. Also, another very important function of the Connector is that it maps the underlying source's data model into relational entities (like tables and procedures etc). So, naturally when you are dealing with RDBMS sources they represent the sources underlaying data model. However, if a connector represents a CSV file, it exposes the metadata information about csv data in "relational" terms to the out bound systems, thus a CSV file looks and behaves like a RDBMS source to a calling system. Using this technique can can virtually transform any source and represent it in relataional terms. Thus you can even call Teiid as "Relational Data Abstraction" engine. So, if you have non standard source of data you can map that into relational terms and use it as RDBMS. Now, as you can imagine, magically all the sources under Teiid system look and behave like relational sources. Now on to integration of these sources.

Query Engine: Now let's assume, a Teiid system is setup with two separate connectors. When a end user application submits a SQL based query, the Query Engine uses the metadata/schema of the logical model to parse, validate, optimize and split the query into multiple source specific queries. The query engine will parallelize the source access if possible and process the data into the final result. The query engine has numerous optimizations. The most important one is that it pushes the maximum amount of work to the source connector, thus avoiding any expensive in-memory operations.

End user applications can access the Teiid mainly using JDBC API. That means you could use Hibernate easily and never see any difference. With some additional tooling you can even access your data as Web Service. Teiid project provides tools for every aspect of the development and deployment of model driven data integration.

Before I end this article, there is one very important concept I want leave with and that is Teiid is not a ETL or Data Mart tool, where the data is duplicated from the original source then served to target systems. Teiid works with data from their original locations as it is, there is no unnecessary duplication/coping of data.

Now that hopefully you know "What Teiid really is", you can see how it can solve your company's data integration needs easily. Let us know how we can help.