How to Review EII Software

Another guest post by Peter Spielvogel, Director of Product Management here at Ipedo - T.M.

It seems that our customers have gotten much more sophisticated in their testing methods over the past year. Rather than relying on magazine reviews, they are performing their own evaluations, based on rigorous test criteria. These evaluations routinely extend beyond 30 days and include detailed analyses of performance numbers and scrutiny of optimization methods, query plans, and cost statistics. I believe that several factors are driving this phenomenon:

1. Every organization’s data integration needs are unique and each wants to test its own specific use cases.

2. EII technology is becoming better understood, especially how it fits with other data integration approaches, such as ETL and EAI.

3. Information on EII, while increasingly plentiful, does not have enough details to enable a purchase or even evaluation decision.

4. Product reviews by magazine or on-line publications do not go into enough depth or they test to a “lowest common denominator” integration scenario.

Web sites or magazine articles are a great starting point for research on EII. They typically contain detailed information on products, including features, specifications, and comparison checklists. But, few media outlets actually put the software through its paces in a realistic test scenario. So, if you want a reliable and credible side-by-side comparison of EII products, you may get the best results if you do it yourself.

For Enterprise Information Integration (EII) software, a realistic test would involve building views across several different relational sources, a few Web Services, and perhaps some unstructured content. It would examine cost-based optimization techniques, dissect query plans, and keep tight tabs on performance. It would verify that every application that needs integrated data could connect to the EII server, see a single virtual database (ideally in either relational or XML format), and gain access to all the disparate back-end sources mentioned above.

The Data Warehousing Institute (TDWI) recommends focusing on the three key aspects of EII: 

“1. Tool for designing views: This is where a technical user defines custom views of distributed data, making it all look as if it resides in a single database. Applications or business intelligence tools access data through a view (which is also a run-time platform, not just a design-time tool), greatly simplifying the querying and joining of distributed data.

“2. Data integration engine: The engine handles connections to data sources (usually through ODBC or JDBC, less often through native gateways or APIs), as well as data transformation and data quality functions. 

“3. Optimizer for distributed queries: The engine must include a query optimizer that specializes in distributed queries. For the sake of performance, optimization may push processing into source databases, cache data in server memory, or persist data to disk. Due to optimization, most EII products can return a joined result set in near real-time.”

Ipedo offers a good primer on query optimization in its technology blog.

The bottom line is that if you want to verify the best EII product to meet your needs, you will need to test the products yourself. Relying on a product review that uses someone else’s use case in a different test environment will only result in disappointment. If you need help creating a robust test plan that pushes EII product through their paces in a reasonable time frame, feel free to contact me.