Looking inside

Enterprise search offers companies several options to deal with data overload

See correction below

IT SOUNDS SO SIMPLE: Good, basic enterprise tools that allow search-weary workers to hunt for — and actually find — pertinent data lurking within their own corporate repositories.

Yet the choices now available to improve search functionality and simultaneously curb infosprawl within the enterprise are almost as dizzying as the overabundance of data that sends IT managers scouring for solutions in the first place.

Oh, the decisions that must be made. Whether to choose one of the Internet newcomers honing in on the enterprise market. Whether to sink resources into complicated classification systems. Whether and how to weave search into portals or workflow plans.

Indecision, it seems, is no longer an option. “Highly paid people are spending too much time going through stuff,” notes Hadley Reynolds, an analyst at Boston-based AMR Research.

In fact, enterprise employees may lose as many as three hours per day on often-futile information hunts, causing the top 1,000 U.S. companies to fritter away $2.5 billion annually, concludes market research firm IDC. But stopping search-related hemorrhaging of time and money can be difficult.

“It’s harder than you think. You have to embed search into an infrastructure, so you have to have an idea what you are embedding into before you pick a solution,” attests search customer Larry Quinlan, CIO of Deloitte Consulting in New York.

Deloitte tapped search vendor Verity to build into its Broadvision portal the ability for employees to search not just across the enterprise but across various applications. Making sure the end product was a practical tool was paramount. “We wanted to make sure it was not just an academic exercise,” Quinlan says.

Because enterprise search is more involved than keyword searches and winds through all parts of an infrastructure, most large companies are loathe to farm out internal search capabilities.

Still, in-house solutions present a somewhat costly enterprise investment. Relying on sophisticated search engines and complex algorithms, enterprise solutions promise to hunt information by exploring various corporate repositories, regardless of format. This is commonly called the “discover” phase of search.

Next, the tools sift both the hits generated in the discovery phase and accompanying metadata and present results to users as a single query.

They do vary by vendor, but search engines are now usually bundled with ready-made classification systems that automatically index a company’s digital assets. Often included as well are ready-to-build “taxonomies” that let companies further tailor classification methods to match business processes. But taxonomies can be taxing, and many enterprises turn cold on the idea of training systems to know what to do with documents and then training people to use taxonomies.

Unlocking information

Naturally, a pack of vendors is swooping in to outfit large companies with sophisticated searching tools.

Among them are enterprise mainstays including Verity, which many acknowledge as currently having the largest installed user base. Convera, a multimedia search company formed last September by Intel and Excalibur Technologies, and IBM’s Lotus are also in the mix, although some search companies view Lotus as a potential partner but not a competitor.

Others hustling into the space are Autonomy and Semio, newer vendors with technology tailored for the enterprise. Some, such as Autonomy, balk at the search label: Autonomy describes itself as an infrastructure software company specializing in advanced pattern-matching and neural-network technologies.

In the race for enterprise search accounts, companies such as Verity and Lotus claim their strengths lie in the capability of accommodating different document types and security mandates.

Verity executives estimate that within an enterprise, there are more than 250 formats to be reckoned with and repositories that span an enterprise’s ERP, CRM, and other systems. That’s a far cry from Web searches of HTML documents.

“A lot of things you and I would think of as extinct, large enterprises are still using,” says Verity CTO Prabhakar Raghavan.

Not only must corporate search solutions be able to navigate a plethora of document types, including XML documents, they must do so using appropriate security measures.

Consider documents on pending layoffs, employee performance reviews, or even product road maps; trouble ensues if such documents turn up on the wrong employee’s search hit list.

“In the enterprise, as opposed to on the Web, we are not all equal citizens,” Raghavan says.

Internet-bred enterprise search engine companies acknowledge that security is a big differentiator between the enterprise and Internet search markets. But vendors such as Inktomi and Northern Light claim their Net experience has led to technology that performs and scales to meet enterprises’ demands.

“Most Internet search companies never had to limit based on who the user is,” says David Seuss, Northern Light’s CEO.

But Northern Light claims it is an exception because the company has long had security and authentication built into its back end to manage “premium” search subscribers. Other former Internet players have fired back with security-related partnerships, such as the one between Inktomi and Netegrity.

Still, the pronounced difference in the markets is undeniable.

“It is really a challenge for [Internet companies] to understand the particular needs of the enterprise,” says search customer Art Rangel, senior marketing manager at Cisco in San Jose, Calif.

The learning curve is not necessarily insurmountable, Rangel adds. Cisco hired Northern Light to devise a unified way to give employees access to costly market research reports purchased from a variety of analyst firms. The networking giant counts on the search tool for usage statistics on the subscription services.

Separately, Cisco is investing in search in its move to reposition the corporate Web site. Instead of serving as mostly a “sales engine,” the revamped site will give Cisco employees and others access to a huge array of information based on their identity, Rangel says.

Military precision

Key to the evolution of enterprise search is the ability to tailor solutions to a customer’s specific needs. The U.S. Army, for example, recently sought a search solution that would enable 710,000 users to hunt for information, some of it classified.

Scalability topped the Army’s criteria for search functionality to underpin its new portal. “This was really a non-negotiable,” declares Col. Robert Coxe, director of the U.S. Army Strategic and Advanced Computing Center in Washington.

The Army wanted its numerous users to be able to scour the .mil domain and turn up even documents still in progress. Another requirement: the ability for users to delve into areas cordoned off for particular audiences, such as an LDAP directory.

What’s more, the agency wanted a way for users to post perennial questions and receive current answers, as well as interactive options once a search found relevant information.

“A lot of times agents are asking ‘what if’ questions,” Coxe says. “We wanted the tool to suggest to the user, ‘Look at Sally or Joe’s question, which was similar to yours.’ The theory is that that user could then use instant messaging to [ask Sally or Joe] ‘Did you consider this?’ ”

Given these different criteria and the portal’s scope, the Army — like many of its corporate enterprise counterparts — settled on a multivendor approach to search. Blended in the agency’s solution is Autonomy, Verity, Inktomi, and even Google.

“If you have a substantial intranet site, trying to get a tool that is one-size-fits-all doesn’t quite do it,” Coxe says. “To some extent we bought them all, since they come with different strengths and weaknesses.”

Anchoring search to corporate portals seems a natural next step. Search has evolved alongside corporate intranet and portal efforts, which are no longer dumping grounds for company newsletters, but are emerging as platforms for collaboration.

In fact, Microsoft recently developed its own search engine for its SharePoint Portal offering, positioning search as central to a portal.

Some search vendors, however, report declining interest in outfitting companywide portals with heavy-hitting search tools. “We see a few customers that want holistic portal offerings, but we see more customers that really just want to bring together the pieces they have internally,” notes Verity’s Derek van Bronkhorst, senior vice president of marketing.

The U.S. Army’s Coxe advises businesses to beware portal vendors and others bearing search: “The first thing a contractor is going to want to do is sell you a search engine. It’s low-hanging fruit.”

Instead of taking the bait, buyers should do their homework and make sure they know exactly the tasks an enterprise search solution needs to tackle to be useful for their company.

In the end, Deloitte’s Quinlan’s words ring true: Enterprises must really know what it is that needs searching before deciding on a solution. “[A good search solution] requires different people in the business amid various practices to know how different documents will be used as they are shared,” he adds.

020107fesearch.gif

Correction

In this article, we misreported the annual amount companies lose due to often-futile information hunts. An enterprise employing 1,000 knowledge workers would waste nearly $2.5 million per year. We also misreported Hadley Reynolds’ title and the company he works at. He is director of research at Delphi Group.

Source: www.infoworld.com