Search Engine Indexing Process

There is a lot of speculation about how search engines index websites. The topic is shrouded in mystery about exact working of search engine indexing process since most search engines offer limited information about how they architect the indexing process. Webmasters get some clues by checking their log reports about the crawler visits but are unaware of how the indexing happens or which pages of their website were really crawled.

While the speculation about search engine indexing process may continue, here is a theory, based on experience, research and clues, about how they may be going about indexing 8 to 10 billion web pages even so often or the reason why there is a delay in showing up newly added pages in their index. This discussion is centered around Google, but we believe that most popular search engines like Yahoo and MSN follow a similar pattern.

Google runs from about 10 Internet Data Centers (IDCs), each having 1000 to 2000 Pentium-3 or Pentium-4 servers running Linux OS.

Google has over 200 (some think over 1000) crawlers / bots scanning the web each day. These do not necessarily follow an exclusive pattern, which means different crawlers may visit the same site on the same day, not knowing other crawlers have been there before. This is what probably gives a daily visit record in your traffic log reports, keeping web masters very happy about their frequent visits.

Some crawlers jobs are only to grab new URLs (lets call them URL Grabbers for convenience). The URL grabbers grab links & URLs they detects on various websites (including links pointing to your site) and old/new URLs it detects on your site. They also capture the date stamp of files when they visit your website, so that they can identify new content or updated content pages. The URL grabbers respect your robots.txt file & Robots META tag so that they can include/exclude URLs you want/do not want indexed. (Note: same URL with different session IDs is recorded as different unique URLs. For this reason, session ID's are best avoided, otherwise they can be misled as duplicate content.) The URL grabbers spend very little time & bandwidth on your website, since their job is rather simple. However, just so you know, they need to scan 8 to 10 Billion URLs on the web each month. Not a petty job in itself, even for 1000 crawlers.

The URL grabbers write the captured URLs with their date stamps and other status in a Master URL List so that these can be deep-indexed by other special crawlers.

The master list is then processed and classified somewhat like:

• New URLs detected
• Old URLs with new date stamp
• 301 & 302 redirected URLs
• Old URLs with old date stamp
• 404 error URLs
• Other URLs

The real indexing is done by (what were calling) Deep Crawlers. A deep crawler's job is to pick up URLs from the master list and deep crawl each URL and capture all the content - text, HTML, images, flash etc.

Priority is given to "Old URLs with new date stamp" as they relate to already index but updated content. "301 & 302 redirected URLs" come next in priority followed by "New URLs detected". High priority is given to URLs whose links appear on several other sites. These are classified as important URLs. Sites and URLs whose date stamp and content changes on a daily or hourly basis are stamped as News sites which are indexed hourly or even on minute-by-minute basis.

Indexing of “Old URLs with old date stamp” and "404 error URLs" are altogether ignored. There is no point wasting resources indexing “Old URLs with old date stamp”, since the search engine already has the content indexed, which is not yet updated. "404 error URLs" are URLs collected from various sites but are broken links or error pages. These URLs do not show any content on them.

The Other URLs may contain URLs which are dynamic URLs, have session IDs, PDF documents, Word documents, PowerPoint presentations, Multimedia files etc. Google needs to further process these and assess which ones are worth indexing and to what depth. It perhaps allocates indexing task of these to Special Crawlers.

When Google schedules the Deep Crawlers to index New URLs and 301 & 302 redirected URLs, just the URLs (not the descriptions) start appearing in search engines result pages when you run the search "site:www.domain.com" in Google. These are called supplemental results, which mean that Deep Crawlers shall index the content soon when the crawlers get the time to do so.

Since Deep Crawlers need to crawl Billions of web pages each month, they take as many as 4 to 8 weeks to index even updated content. New URL's may take longer to index.

Once the Deep Crawlers index the content, it goes into their originating IDCs. Content is then processed, sorted and replicated (synchronized) to the rest of the IDCs. A few years back, when the data size was manageable, this data synchronization used to happen once a month, lasting for 5 days, called Google Dance. Nowadays, the data synchronization happens constantly, which some people call Everflux.

When you hit www.google.com from your browser, you can land at any of their 10 IDCs depending upon their speed and availability. Since the data at any given time is slightly different at each IDC, you may get different results at different times or on repeated searches of the same term (Google Dance).

Bottom line is that one needs to wait for as long as 8 to 12 weeks, to see full indexing in Google. One should consider this as cooking time in Googles kitchen. Unless you can increase the importance of your web pages by getting several incoming links from good sites, there is no way to speed up the indexing process, unless you personally know Sergey Brin & Larry Page, and have a significant influence over them.

Dynamic URLs may take longer to index (sometimes they do not get indexed at all) since even a small data can create unlimited URLs, which can clutter Google index with duplicate content.

What to do

Ensure that you have cleared all roadblocks for crawlers and they can freely visit your site and capture all URLs. Help crawlers by creating good interlinking and sitemaps on your website.

Get lots of good incoming links to your pages from other websites to improve the importance of your web pages. There is no special need to submit your website to search engines. Links to your website on other websites are sufficient.