Intelligent Enterprise

Better Insight for Business Decisions

RSS
Webcasts
Whitepapers
Subscribe
Home

January 1, 2002

/020101/501decision1_1.jhtml">

Out in the Open

In the world of analytics, the open-source torrent is little more than a trickle

By Seth Grimes

Continued from Page 1

Comshare Inc.'s now-defunct System W, which didn't survive IT's move from mainframe timesharing to the client/server world in the '90s, is an exception that proves the rule. Perhaps Comshare was worried about protecting its intellectual property — its investments into development of highly specialized analytic data structures and algorithms — even as it was abandoning its product. Or maybe Comshare's perception matched what John Kopcke, CTO of BI vendor Hyperion Solutions Corp., told me: Users would be concerned about the reliability and robustness of OS products and that the BI market is simply too volatile to support OS development. That volatility may be the reason noncommercial developers seem not to be tackling mainstream BI problems like OLAP slice-and-dice analysis.

Lastly, the continued popularity of spreadsheets as analytic interfaces discourages the emergence of OS analytic clients: Spreadsheet products are ubiquitous so there's no added purchase cost.

What could jump-start OS decision support, particularly in mainstream BI?

COMPETITIVE STRATEGY

First, the example of Borland Software Corp.'s release of InterBase to the OS community illustrates that a tool vendor that found competing for sales no longer profitable — I can think of several qualifiers in the online analytical processing (OLAP) market — could open its code base. This scenario isn't far-fetched. Microsoft is creating price pressure on the low- to mid-range OLAP server market, and the integration of analytic capabilities into mainstream RDBMSs may prompt some shops to drop specialized software in favor of jack-of-all-trades servers.

But creating data warehouses and data marts and designing analytic client applications still require specialized business and programming skills. Despite analysts' best efforts — Gartner got this one right early — the larger software market still seems to think wrongly that low-cost, commodity products necessarily have low lifetime total cost of ownership (TCO). This misperception implies that making a good product available free should boost that product's market share in the short term, expanding potential lucrative consulting opportunities. In the longer term, a move to open source could lead to developers implementing new analytic algorithms and techniques and new interfaces at no research and development cost to the (former) vendor.

Or an OS release could be part of a larger, overall competitive strategy. I'm thinking in particular of Oracle, which states that Oracle 9i OLAP, integrated into the company's flagship DBMS product, "is a simpler, faster problem-solving approach than using piecemeal tools and applications," presumably including Oracle Express. I could see Oracle releasing Express source code in an effort to damage low- to mid-range OLAP competitors like Cognos Inc. and Microsoft.

Rate This Article

Comments:

Optional e-mail address:

Will we see OS analytic client tools before we see OS analytic DBMSs? It's a lot easier to develop a client tool than a database server, and quite a few consulting companies have transformed tailor-made interfaces built for particular customers into general-purpose tools for the wider market. But rapid, low-cost development is eased by widespread use of commercial interface objects — data grids, graphics and visualization functions, and so on — written especially in Java, C++, and Visual Basic, and packages as applets or controls. Reliance on third-party code is an obstacle to source-code release. In principle, a developer could use OS utilities like gnuplot, but I see little awareness of those utilities to date. At least proprietary interfaces won't stand in the way of this development given the emergence of open APIs.

John Kopcke's thought — that support for OS efforts is discouraged by the BI market's volatility — may explain best why OS decision-support projects have been limited to specialized statistical analysis. But volatility creates opportunities, and I'm hopeful that the mainstream BI world will be cracked open soon.

Seth Grimes [grimes@altaplana.com] is a principal of Alta Plana Corp., a Washington, D.C.-based consultancy specializing in analytic computing systems and demographic and economic statistics.