Google Search Appliance indexes 10 million documents with one box

Upgrade also brings personalized search features

An upgrade to the Google Search Appliance for businesses allows one box to index 10 million documents, more than triple the previous amount, and offers new end-user features including personalized alerts and results ranking.

The hardware-and-software bundle, based on Dell servers, sits behind a customer’s firewall and crawls content in corporate intranets, Web servers, portals, file shares, databases, content management systems and various business applications.

Previously, one box could crawl up to 3 million documents. While the appliance’s minimum capacity is 500,000 documents, many customers’ search needs extend beyond 10 million, says Google lead enterprise search product manager Nitin Mangtani.

To continue reading, register here to become an Insider. You'll get free access to premium content from CIO, Computerworld, CSO, InfoWorld, and Network World. See more Insider content or sign in.

An upgrade to the Google Search Appliance for businesses allows one box to index 10 million documents, more than triple the previous amount, and offers new end-user features including personalized alerts and results ranking.

The hardware-and-software bundle, based on Dell servers, sits behind a customer’s firewall and crawls content in corporate intranets, Web servers, portals, file shares, databases, content management systems and various business applications.

Previously, one box could crawl up to 3 million documents. While the appliance’s minimum capacity is 500,000 documents, many customers’ search needs extend beyond 10 million, says Google lead enterprise search product manager Nitin Mangtani.

Click to see: Google Search Appliance

With competing software-based search products from vendors like Autonomy and the Microsoft subsidiary Fast, an enterprise would need several pieces of hardware: a Web server, an application server and database server for indexing, says Enterprise Strategy Group analyst Brian Babineau.

“It’s literally like managing a mini-data center if you have to scale to 10 million documents [with other search products],” Mangtani argues. The upgraded Google appliance, announced Wednesday, will be available at the beginning of September.

While 10 million documents on one box sounds impressive, there’s one caveat with the Google appliance, Babineau notes. It represents a single point of failure. If you require clustering and high availability, you’ll need to buy two, he says.

Standard configurations of Google’s Search Appliance allow indexing of up to 30 million documents, a number unchanged from previous versions. Indexing 30 million documents requires 12 servers in the upgraded version, just as it did in previous editions, because extra hardware is needed for high availability and backup.

When Forrester Research analyzed the enterprise search market in May, the firm said Google competes effectively with ease of use and low prices, but that the most impressive technical capabilities are found in rival products from Autonomy, Fast, Endeca and Vivisimo.

Google says its new release adds several new features improving the experience for end users and systems administrators. A new alerts system lets employees receive e-mails when there are new documents related to topics they are interested in, similar to how the publicly available Google Alerts system provides updates from Web-based content.

Administrators can now bias results based on metadata, in addition to results biasing based on the author of a document or date it was created. Administrators can also adjust search results based on the department a person works in or job function. For example, “engineers might choose to weigh code or design documents higher, while marketers might prefer marketing documents,” Google says.

More upgrades include spell checking in Portuguese, French, Italian, German, Spanish and Dutch, and support for the Kerberos (protocol) network authentication protocol.

The Google Search Appliance starts at $30,000, including two years of support, for the entry-level version that can index 500,000 documents. Google said it would not publicly disclose pricing for the more robust versions, including the single box that indexes 10 million documents.