Press Release / Web sites / Internet / Google Search Appliance vs Arch

Google Search Appliance vs Arch

We compare Google Search Appliance with the enterprise search engine Arch (http://www.atnf.csiro.au/computing/software/arch/) that we developed at CSIRO and license under a CSIRO Open Source software licence, and argue that Arch can be a good replacement for GSA. This comparison covers the essential criteria that influence the cost of the solution vs its usefulness, i.e. value.

- Scalability: both Arch and GSA can work on clusters of computers and offer unlimited scale. The difference is in the price you pay. Arch is free, GSA is $32K - just for one node.

- Cost of deployment and maintenance: both are easy to deploy and maintain, and offer almost a “turnkey” solution in simple cases. We discussed this topic in article “Enterprise Search Engine in 15 Minutes?” (http://www.atnf.csiro.au/computing/software/arch/ArchWebArticle3.pdf)

- Query power: GSA supports wildcard searches, spelling correction and ordering on a set of document attributes. Arch offers full power of Apache Solr and very powerful Lucene query syntax.

- Supported types of indexed documents: both GSA and Arch offer a set of parsers that cover all common document formats.

- Supported types of document sources: Both GSA and Arch are able to index non-web data, such as the contents of relational databases. Apache ManifoldCF is a connector framework providing Solr connectors that let Arch index data residing in enterprise data repositories, such as FileNet P8, Documentum, LiveLink, Meridio, Windows Shares, SharePoint, relational databases and others.

- Index completeness: With web log processing enabled, Arch is able to provide a more complete index than GSA by finding isolated web pages that "normal" web crawling algorithms, including those used by GSA, will not find.

- Security: both products support document level access control. Arch also supports an unlimited number of secure search gateways that can serve pre-filtered queries to narrow search for security or relevance reasons.

- Flexibility: both products have clearly defined APIs and extension points, but, being an open source software package, Arch is more modifiable, extendable, and therefore more flexible, able to accommodate virtually any custom requirements.

- Relevance of results: arguably, this is the most important criterion that makes a difference between success and failure of the search engine. See "Corporate Search: Can We Just Get Google?" (http://www.atnf.csiro.au/computing/software/arch/ArchWebArticle.pdf). On a set of 47 test queries and over 100K documents, Arch overperformed GSA by about 10% on average.

Arch and GSA are comparable by the criteria above. However, being open source and thus more flexible, Arch may provide a solution in some cases where GSA can’t. As Arch is free, flexible, and provides at least comparable to GSA performance in relevance, the most important quality criterion for a search engine, it clearly represents much better value for money for most use cases.