分页大量数据

Question

Something is very common results of a search, for example, paging, and is usually shown the total page, results, etc ...

Two examples:

Google - https://www.google.com.br/?gfe_rd=cr&ei=h4PzVLy3JMiU8QeE3ICYBQ&gws_rd=ssl#safe=off&q=test - approximately 2.67 billion results (0.39 seconds)

Stackoverflow - https://stackoverflow.com/search?q=test - 1,561,211 results

For example, Google, assumes that these 2.67 billion results are found by the database to be counted. As is done to count the total results of a query with many results without compromising performance? What magic is this?

Answer 1

"approximately 2.67 billion results (0.39 seconds)" -- Notice the word "approximately". That opens the door wide open to tricks that allow it to do the count in virtually 0.00 seconds. Examples:

• Do some counts every night
• Statistically sample
• Have summary tables

"1,561,211 results", on the other hand 'sounds' like an exact count. But it may be a "stale" result. I refreshed that page a few times; the value did not change. This leads me to believe that they really count it, but then cache the result for some time. Memcached is a likely suspect for such caching.

(There are doubtless other tricks of the trade.)