By Krishna Roy
When Oracle Corp acquired British metadata management specialist, One Meaning Inc two weeks ago, it talked up the acquisition as the strategic purchase destined to put it ahead of Platinum Technology Inc and Microsoft Corp in the datawarehousing league table. Microsoft (with Plato) is not going to be a significant player in the enterprise space and will only make an impression at the low-end of the market. We are stronger than Platinum because they have two repositories and we have one, said Michael Howard, VP of Oracle’s DataWarehousing program office. Such bullish confidence from Oracle is really nothing new, particularly when it comes to the company’s relationship with Microsoft. Oracle, after all, has just launched the latest release of its database, Oracle 8i and is taking every opportunity to blunt the release of SQL Server 7.0 from Microsoft. CEO, Larry Ellison, is as usual taking every public opportunity to rubbish Redmond. But what is surprising is that Howard is talking of a world firmly entrenched in client/server computing, when Ellison has been evangelizing the merits of the internet computing model for some three years. Datawarehousing, after all, relies on the premise that there is a host of data in databases on disparate servers that needs to be pulled together, then queried and analyzed using (usually) fat-client software. All the while Ellison has been advocating the removal of desktop applications onto the server, with the browser left as the only software required on the client. Ellison nevertheless took the opportunity to disparage Plato in the light of the One Meaning acquisition. We are leagues ahead in the metadata library. Ours is real, Microsoft’s isn’t, he told journalists at Oracle OpenWorld. His comments refer to the metadata standard recently submitted to the Object Management Group as part of its Common Warehouse MetaData (CWM) initiative, which both IBM Corp and Unisys Corp are behind. Microsoft is developing its own metadata standard. One Meaning has spent the last nine years honing a metadata (data about data) model, so the acquisition clearly fits in with the initiative.
Addressing inconsistencies
One Meaning’s first tool, when the company was still named Software One Ltd, was called Exchange and was originally developed to address the inconsistencies between data generated by various tools prior to building an application. Exchange sought to remove incompatibilities by mapping metadata between proprietary repositories. The software was licensed to IBM, Unisys and Groupe Bull for their respective repositories, while Oracle developed an interface to Exchange as part of its datawarehousing suite. The company went on to develop Marlow, essentially a warehouse for metadata that uses Exchange to import information from a variety of sources which than can be warehoused into a database known as a metabase. Marlow acts an information store for metadata produced from disparate tools, applications and databases. The metabase tracks versions; the history of the metadata over time; manages access and control to the metadata; and stores the context of the data. As One Meaning began to play an increasingly influential role in the Common Warehouse MetaData initiative, Oracle decided it was time to purchase the company. It wasn’t the only one. IBM, Unisys, Bull, and possibly Sterling Software were among other interested parties. One Meaning was ready to relinquish its independence, having failed to raise a second round of venture capital funding. Its search for funding continued unsuccessfully until the end of August when serious negotiations with Oracle began. Why were no investors interested? Perhaps because the pace of ERP and datawarehouse implementations is perceived to be slowing in the face of Year 2000 budget re-allocation. One Meaning’s products only address a specific piece of the datawarehousing puzzle – the capture and storing of metadata – and many of its larger competitors provide tools to build full datawarehouses. Not only that, the initial venture capital was raised on the perceived success Marlow would have in the market. But as Tom Bredt from Menlo Ventures told us: We raised money to deploy Marlow but the sales cycle was longer than we had anticipated and we didn’t attract as many customers as we had hoped. Marlow required a direct sales force and we didn’t have enough time or money to build one. The transition from the UK to the US also took more time than we had anticipated. The operational aspects of One Meaning ( It was run like an R&D shop rather than a business destined to make money, we were told) won’t be of concern to Oracle, which bought the company for its metadata expertise and some 24 technicians. Effectively, Oracle has bought metadata technology to fill out its repository offering that is currently based on its homegrown CASE tool, Designer/2000, regarded as good for application development, but not so effective for datawarehousing. Exchange includes bridges to 26 warehouse tools and applications that will enable Oracle to support a range of third-party data warehouse environments. Exchange and Marlow will appear in Warehouse Builder 2.0, a suite approach for all aspects of designing, creating and managing a datawarehouse application that will share a common metadata management tool under CWM. The suite, which uses Oracle 8i for its database engine, goes into beta this December and will not be released until the second quarter of 1999. It will effectively tie together Oracle’s Discovery decision support and Express multi-dimensional tools. Marlow and Exchange will be used to extract, transfer and publish metadata to the Oracle tools. Marlow is currently built around Versant Corp’s object database, and the close tie has prompted speculation that Oracle might also buy the object database supplier – something which Oracle denies. It argues that it is relatively easy to separate the Marlow application layer from the object database and that it will build its own application layer to Oracle 8i using Unisys Corp’s URep repository. The development will occur within the next eight months and promises more data integrity features that those currently available. One Meaning has an installed base of around 400 customers, the majority of which use Exchange. Marlow, which is just over a year old, is used only in a handful of sites, notably ING Barings. Oracle has said that it will continue to maintain both point products, although eventually it is likely to want to migrate sites over to the new Warehouse Builder product. Oracle will work with half a dozen Marlow customers over the next eight months or so to ensure that they complete their development projects and then, presumably, urge them to use Warehouse Builder for forthcoming projects. Marlow customers are highly visible and at a critical stage in their development projects, so we will work with them very closely, Howard told us. Oracle’s strategy is to sell the datawarehousing packaged application suite into its existing installed base – a sensible decision given that it will generate a steady flow of revenues. While that revenue stream is difficult to quantify, Oracle has described it as ‘huge’ and has made some upbeat estimates on the growth of its applications business as a whole that have caused The Street to revise earlier estimates. Merill Lynch, for example, has doubled the revenue growth expectation for Oracle’s applications business on the basis of management indications that it will see 30%-40% application license growth in 1999.
