EII Meets BI

With its ability to easily pull together disparate data on the fly, an emerging class of software called Enterprise Information Integration (EII) is being touted as a flexible alternative to traditional data warehousing.

EII software targets the age-old challenge of enterprise data integration and analysis. The technology works by providing unified views of multiple heterogeneous data through a distributed, or federated, query. EII is radically different to other data integration technologies because it neither moves data nor creates new stores of integrated data. It leaves data where it is, leveraging metadata repositories across multiple back-end systems to pull information transparently into new applications.

But can EII live up to its potential billing? Some industry analysts warn that companies pushing EII initiatives need to spell out in greater detail how it is going to be managed as deployments and queries scale up and across the enterprise.

What is EII?

The best way to think of EII is as a database 'veneer' that allows users to access and query data as though it is residing in a single database. EII takes an existing database capability that federates queries across different tables, and implements it on a virtual basis. EII makes numerous data sources look like one system, even though behind the scenes there are separate systems with their own code and engines.

Despite its cost and complexity, the rewards of EII can be great. The most obvious benefit is for business intelligence reporting. EII's ability to provide timely

BI without having to retrofit or adapt a data warehouse for specific decision-support tasks is referred to by EII vendors as 'on-demand intelligence'.

Vendors cite a range of useful BI applications using this approach: operational dashboards that track functional performance; financial risk analysis, where each single transaction might effect a change; and retail operations, where hourly metrics can guide selling and supply strategies.

It is natural for BI vendors to look to EII to glean operational intelligence from non-traditional data sources outside the scope of most corporate data warehouses. Recent moves by EII vendors to align themselves with BI confirm the ties between the two. Composite inked a deal with Cognos to build its EII tools into Cognos' ReportNet query and reporting software. MetaMatrix has forged technical partnerships with Business Objects and Hyperion Solutions. And Actuate bought EII specialist Nimble Technology to integrate data query federation capabilities into its enterprise-reporting platform.

EII's ability to pull in data and create unified views quickly and cheaply is also a key selling point in BI environments. The argument is simple: why spend time and money building a redundant data store when you can create a federated query to grab the data when you need it?

Proponents argue that EII replaces a physical extract of a data warehouse, thereby removing the need to spin off expensive data marts. They have a point. After all, what is the point of integrating a large amount of data when you only need a small slice of it?

Some EII vendors have been bold enough to suggest that EII could eventually replace traditional data warehousing. They might be excused for their early exuberance, but does EII really compete with data warehousing? The smarter EII vendors now guard against this line of thinking and articulate the complementary positioning of both technologies. They are right.

EII should not be seen as a direct replacement for a data warehouse. Rather, it augments historical time-series BI reporting with fresher operational detail. It can also federate data from the warehouse itself, and present it to the user or client application on demand.

EII is especially useful in situations where you want to get to data that is omitted from the warehouse or in situations where you simply cannot get at the data through standard query languages. Bolting on an EII capability allows you to supplement data warehouses with lighter queries running directly against current, or intra-day transactional data from operational systems. Equally, EII is not a replacement for deep, complex analytic processing. It is not the right tool for multi-terabyte queries: the data warehouse is.

Great expectations

As with all emerging software categories, the larger business intelligence vendors are expected to partner with EII companies, build or acquire it themselves. Until this happens, the market will remain highly fragmented.

From a pure BI perspective, it is fair to say that there is a very restrictive market for EII that leans towards a subset of BI queries that requires data to be extraordinarily real-time and up to date. Today most fundamental BI queries simply are not of that type and do not require data to be even remotely real time.

So while EII is still a long way from mainstream adoption, it is close enough to inspire excitement in anticipation of its arrival.

POTENTIAL DRAWBACKS OF EII

* The risk of over-stressing operational systems: EII queries can be unpredictable and their impact on operational systems can, too.

* Lack of trending: the EII approach inherently means there is no history.

* Few safeguards of data quality: with EII, data quality is more dependent of the integrity of the source systems.

SOURCE: CBR