Next: Florida-Update
Previous: Entry Page

Everflux

Traditionally Google makes one major update per month. This is dubbed the Googledance as rankings in the results pages change erratically, particularly between the various data centers that serve Google's results.

This is then followed by visits from the Googlebots, the results of which are fed into the next dance. Since around mid-2002 webmasters noticed that sites were being spidered on a more continuous basis. This crawl is an attempt to identify new or updated content in order to keep the Google index fresh. It is neither as broad nor as deep as the regular indexing process. It can also produce some strange effects where new sites are located by an inbound-link or submission to Google. Some content gets indexed and shows up in the results pages only to be dropped, sometimes for a couple of months before the main crawl indexes the site.

When a user goes to www.google.com they are in fact being routed to one of thousands of computers that handle queries. The routing is handled by load-balancing, either by the Internet naming system (DNS) returning different addresses for the domain or by front-end techniques entirely transparent to the end user. Google's computers are built out of cheap, off-the-shelf components running the Linux operating system and are mainly located in data centers in California and Virginia. Google is thought to have a number of separate copies of its index located across these computers. The same search can yield slightly different results depending on which copy of the index is used.

The effect of continuously fluctuating search results has been dubbed Everflux by search engine experts.

Everflux

See Also