What the Hell is IBM Information Integrator?

It's a database thang

Cat 5 cable

Briefing Note Yesterday IBM announced its new Information Integrator family of products, writes Phil Howard. Ultimately this will consist of three offerings (although in the longer term the three products will probably converge), based on SQL, an object oriented API and an XML API respectively. However, the last of these, which will use XQuery, has not been announced yet, as it is awaiting the final definition of the XQuery standard.

The two products that are announced (in beta) are Information Integrator, previously known by the code name of Xperanto; and Information Integrator for Content, where the former is the relational product and the latter is designed to provide integration in content management environments. In practice, the latter represents a re-positioning of IBM's Enterprise Information Portal (a subset of WebSphere Portal, which is really the company's enterprise information portal) for accessing mainly IBM content repositories together with other content and data sources.

The Content product is not really new, so I want to focus on Information Integrator, which is now available for beta testing and which is scheduled for general availability in mid-summer. In today's article I will discuss the details of Information Integrator and tomorrow I will consider the circumstances under which it will be most appropriate to look at Information Integrator as opposed to alternative technologies such as ETL (extract, transform and load) tools. I will also consider some of the environments in which use of Information Integrator may be most beneficial.

Information Integrator (which is in version 8.1 to align it with the latest release of DB2) is based primarily upon the facilities of DB2, SQL and DataJoiner. The basic concept is predicated upon a federated database approach in which multiple heterogeneous databases appear to the user as if they were a single database.

Both Microsoft and IBM have espoused this approach for some time, while Oracle has preferred to concentrate upon centralisation. However, the downside of centralisation is that you have to rip out and replace existing databases, with all the pain that that entails.

Microsoft, meanwhile, has relatively limited support for federated databases in SQL Server 2000 and, even then, it tends to be limited to SQL Server support, whereas IBM has taken a more agnostic approach, supporting all sorts of relational databases within a federation. It is not hard to say, therefore, that IBM is the market leader in this space.

However, it is also important to realise that Information Integrator is not limited to accessing relational data sources - it can also access XML, flat files, Microsoft Excel, ODBC, Web and other content stores and so on, although updates and replication are limited to relational sources in the first release. Thus (for those of you who know the product) the full capabilities of DataJoiner have not been implemented in this release.

There are some key features of Information Integrator that should be mentioned. In particular, you can query data wherever it resides, as if it was at a single location, with a single view across all the relevant data sources. The product supports queries by caching query tables across federated sources, while the optimiser will validate the SQL used against the source database and will automatically compensate if the relevant syntax is not supported on the remote database. Other features of the federation capabilities of the product include the ability to publish the results of a query to a message queue and to compose, transform and validate XML documents.

In terms of updates, I have already mentioned replication and Information Integrator effectively acts as a replication server, initially supporting Oracle, Informix, Microsoft, Sybase and Teradata databases, as well as DB2. Functions are flexible with support for both one to many and many to one topologies; table-based or transaction-based data movement, which may be dependent on whether you have batch or online requirements; and latency which may be scheduled, interval-based or continuous.

While a brief article such as this can give no more than a flavour of a product like Information Integrator (and Bloor Research will be publishing a full report on the product in due course), it should be clear that in the right environment Information Integrator has much to offer. What those environments might be, I will discuss tomorrow.

© IT-Analysis.com

Sponsored: Designing and building an open ITOA architecture