CNET también está disponible en español.

Ir a español

Don't show this again

Tech Industry

Commentary: IBM, Yahoo looking to leapfrog Google

Google has had an unbreakable grip on entry-level enterprise search for some time now. That's about to change.

Commentary: IBM, Yahoo looking to leapfrog Google
By Forrester Research
Special to CNET
December 13, 2006, 4:40PM PT

By Matthew Brown, with Chris Mines and Jamie Barnett

The latest in IBM's line of enterprise search products, IBM OmniFind Yahoo Edition, is powerful, simple to use and free.

IBM just announced availability of a freely downloadable enterprise search product that installs with three clicks, packs the same features as a Google Mini, scales to half-a-million documents, and is built on open-source technology from the Apache Lucene project. This bodes well for information and knowledge management professionals deploying entry-level search capabilities but poorly for Google Enterprise, Microsoft, Oracle, SAP and smaller vendors still looking to cash in on the entry-level enterprise search market.

related story

Companies to offer a free data-search tool for businesses,
a quirky move challenging Google in a blossoming mark.

Google created the market for basic enterprise search in 2003 with its Google Search Appliance--its focus was not on fancy new features but on powerful, easy-to-use search at a reasonable price. Google moved down-market while incumbent enterprise search vendors moved up--rapidly expanding its line of search appliances to Google Minis and making it simple for companies to buy, install, configure and start searching. Pricing for Minis starts at $1,995 for 50,000 documents and grows to $30,000 (for the full-fledged Google Search Appliance) searching up to 500,000 documents.

Meanwhile, incumbent software platform vendors IBM, Microsoft, Oracle and SAP watch enviously as 1,000 flashy yellow and blue search appliances fly off of Google's shelves every quarter--freckling the drab, gray server rooms of the world's largest enterprises.

The halo Google created around entry-level enterprise search has, to date, been impenetrable. IBM, Microsoft and Oracle have tried to get a piece of the action--with competitively priced and/or more full-featured products. But this is not a feature war; the broader "software platform" value proposition these companies sell has failed to slow Google down thus far. This is about to change. A combined IBM, Lucene and Yahoo will be a different story because:

• Yahoo brings a familiar interface, Internet search, and consumer mind share. Let's be clear: OmniFind Yahoo Edition is not Yahoo's core search engine packed into an installable software package. But it is Yahoo's familiar, simple-to-use interface. And executing searches against enterprise intranets, file systems or Yahoo's public Internet search couldn't be much easier using this product. Yahoo also brings a prominent, direct, Google-like distribution channel--planning to offer the product for download directly from its site--rather than burying it deep within the bowels of IBM's own site.

• Lucene offers a solid, open, full-text indexing core. The Lucene relationship brings all of the benefits of the open source community--rapid innovation, interoperability, and open, standards-based architecture--without the common open source risks like lack of organized professional services, support and complexity. Lucene's core search libraries already offer high-performance batch and incremental indexing, powerful fielded attribute-level searching, relevance biasing and other capabilities--while other open source projects include extensive capabilities around parsing common Microsoft Office document formats, PDFs and others.

• IBM ties it together with simple management and world-class support. Perhaps the most shocking aspect of this product is how easy it is to get up and running--especially coming from IBM. IBM lets customers install the product, point it at a Web site or file system, and start indexing--all within three to five clicks. Of course this simple interface exposes only the most essential capabilities--user interface branding tools, query spell checking, thesaurus tools, best-bets matching (or Keymatch in Google-speak) and basic reports--rather than all of the underlying capabilities of the engine that are accessible through lower-level APIs.

• And it's all free for up to 500,000 documents. The fact that IBM is capping document capacity at 500,000 documents is significant considering all major competitors charge in the mid-to-low five figures for this searchable capacity.

What it means for large enterprises
Information and knowledge management professionals in large enterprises gain on three fronts:

• Finally, an alternative to Google. For the past four years information and knowledge management professionals have been bludgeoned by their business stakeholders demanding "just give me Google" for search. To date they've had few alternatives to consider--other than high-priced products from established vendors like Autonomy, Convera, Endeca and Fast, or upstarts like Coveo, Exalead or Isys Search Software. To date, companies also perceive insurmountable risks and complexities with open source. Now, information and knowledge management professionals will be hard-pressed not to at least consider and try the IBM OmniFind Yahoo Edition product.

• A search product that runs on corporate standard hardware. Architects planning for search in the context of their broader information management strategies will welcome the open, nonproprietary nature of the product--gaining the ability to run it on low-cost, standard, company-approved hardware and enabling them to extend the product to more data sources, file types and enterprise applications than other products offer. Many companies find the combination software/hardware product offering from Google tough to swallow, and several Google partners criticize the fact that key aspects of its search API are still kept under wraps by the company.

• World-class enterprise support programs and nonproprietary open standards. Large IT departments will appreciate IBM's support program behind this product--an area where Google has been weaker. Further, enterprises looking to exploit underused hardware assets and plan for future integrations will appreciate the nonproprietary, extensible nature of the core Lucene indexing engine.

IBM and search
Over the past year and a half, IBM has added substantially to its search product line--acquiring natural language search provider iPhrase, announcing its IBM OmniFind Enterprise and Discovery Editions, and fostering broad-based innovation around the Unstructured Information Management Architecture (UIMA) standard. This most recent announcement will help IBM to extend the OmniFind brand and to offer an on-ramp for those customers coming to enterprise search for the first time--even as the company continues to build out higher-end search products for solutions as diverse as e-commerce and intranets, quality management tools, reputation monitoring and analytics applications for the public sector.

But IBM's strategy is not without risks because:

• The strategy is unproven. It's unproven whether low- to no-cost search products are truly entry points into higher-end product sales. These days simple search indices can be built and rebuilt quickly, and basic search configurations are relatively easy to do. Unlike high-end search implementations--requiring deep integration and custom configurations--buyer-switching costs for basic search are relatively low. Indeed, even Google's partner network reports that entry-level buyers are slow to adopt the advanced capabilities they offer like results clustering, secure connectors, and enterprise application and repository integration.

• A software-only approach will be shunned by "plug-and-play" buyers. Nontechnical buyers that appreciate the all-in-one Google software/hardware product will consider total cost of ownership with OmniFind Yahoo edition--including the time, effort and expense of procuring hardware necessary to install and run the engine.

• The product won't sell itself. While unconventional for enterprise software sales, Google has built awareness and generated much of its growth through telesales professionals actively contacting prospective buyers. IBM will struggle to find sales professionals within its ranks willing to promote this product.

Yet, despite these challenges, of all the strategies put forth by competitors in the past year, this one has the best legs for slowing the Google freight train.

Not just another desktop search tool
This new product is most comparable with Google Minis, but given its hefty capacity, it also takes a jab at products from Microsoft, Oracle and SAP. Don't let the freely downloadable executable fool you into thinking this is just another desktop search tool--this is a real enterprise search to be run on a server--it's just easier to acquire, install and use than just about anything else on the market right now.

• Continue to look to higher-end products for advanced needs. The definition of enterprise search is constantly evolving--as are the capabilities implied by the name. Buyers will find that IBM OmniFind Yahoo Edition serves basic full-text search needs well at a reasonable scale--for retrieving Web pages and documents based on a user's query. But companies that need more advanced capabilities like massive scalability, faceted navigation, clustering or secure, deep connectivity to enterprise apps and data sources, will need to consider IBM's OmniFind Enterprise or Discovery Editions, or high-end alternatives from Autonomy, Convera, Endeca, Fast and others.

• Expect more competition among Google, IBM, Microsoft, Oracle and SAP. From multiple dimensions, this will be the first substantive blow to the Google Enterprise search freight train. Google will likely sustain a strong business for organizations without central IT departments looking for plug-and-play, appliance-based search. But larger IT organizations will find the price, world-class enterprise support programs, and nonproprietary open standards around IBM's product very attractive. For the moment at least, Microsoft, Oracle and SAP will be left with entry-level enterprise search products they cannot easily sell--because, ultimately, it's hard to compete with free.

© 2006, Forrester Research, Inc. All rights reserved. Information is based on best available resources. Opinions reflect judgment at the time and are subject to change.