News 

Although public search from the likes of Google and Yahoo! have increased by leaps and bounds over the past few years, the internal search has lagged behind.

The Google Search Appliance is a combined hardware and software solution. Inside a standard rack mount unit is a standard Intel based server running Linux. On top of which is the Google index and search engines.

Dave Girouard Google's General Manager for Enterprise told us that the company had chosen to provide the service in a combined unit because 'we wanted to make sure that customers got a consistent high quality solution that they would expect from Google.' He also said that by supplying a single rack mount unit, most customers can have the

ADVERTISEMENT

unit up, running and indexing their content in 15 to 20 minutes and start having indexed results within a few hours.

The search engine can understand over 250 file formats, including all the major documents such as Microsoft Office, PDF and of course HTML, and auto detects 28 languages The documents do need to be available via http or https as the Search Appliance views all documents as a unique URL. This means that there are still a very large number of document types held in databases that the Google engine can't index at the moment. 'There is a tremendous amount of legacy content and we aren't there yet,' admitted Girouard but he said Google was 'working on it.'

Documents on a corporate intranet can be organised into an unlimited number of 'search collections' - for example, for particular departments such as sales or product development. Security is seen as a prime consideration and the company says that individuals can be prevented from accessing material they are unauthorised to see.

The basic GB-1001 costs £19,000, which includes two years worth of software support and can index up to 1.5 million documents and handle up to 300 queries per minute. From there, Google offers a range of options up to an eight-way clustered solution able to index up to three million documents.