¡°The Anatomy of a Large-Scale Hypertextual Web Search Engine¡±

	Google¡¯s architecture shares many similarities with existing search engines. It still retains crawler, indexer and searcher. The only new and innovative component in Google¡¯s architecture is the Pagerank module. Unlike past search engines, which rank pages based on frequency of keyword or other similar means, Google uses the Web¡¯s vast link structure as an indicator of an individual page¡¯s usefulness to the browser. In Pagerank, Google interprets a link from page A to page B as a vote from A to B. The weight of the vote depends on the weight of page A itself and the number of links page A contains. In such a ranking system, human tampering or web spam becomes increasingly difficult and users may expect their search results to be more useful.

	Google¡¯s primary goal is to improve the quality of search. However it has done little to address the existing performance issues that all search engines suffer. Google, like other search engines, uses vast numbers of servers to carry out its work. It relies on the decreasing cost of disk storage, memory and other hardware to offset the rising cost of operation and maintenance.

	The sections on Google architecture overview and major data structures have been very informative. I expect this information will help consolidate project ideas if I decide to work on crawler or search engine.