Different titles for deferent queries.

Does Google hold more then one database in a single datacenter?

[dubilbul]

Hi,

I’ve searched Google for a single keyword and I’ve got my site ranked #1 with the NEW title of the site and when I’ve searched for two words phrase (one of the words is the single keywords I’ve searched first) I’ve got my site ranked #98 with the OLD title of the site. First I thought that Google just update its datacenter, but when I’ve searched the first keyword again I’ve got the NEW title once again ranked #1. It is going on for the last 16 hours.
Is it possible that Google holds more then one database on a single datacenter (this can get things complicated with optimizing sites) or maybe it query other datacenters for different queries? (The second option seems to me little bit expansive with transactions)
Is there any one else with such experience?

I’ve found that this subject was discussed before here: [webmasterworld.com...] but then it was a competitor site and the third remained open with questions regarding the competitor technique this time this is my site and I know there is no cloaking or any other tricks.

[Brett_Tabke]

This is the current nature of Google. They are still experimenting with different databases. That makes it appear as though your site is floating around.

...unless of course you are not in the us.

[takagi]

I don't know your site, but sometimes the URL looks identical, but there is a small difference (with or without 'www', an extra '/' at the end, etc). Are you sure they are exactly the same?

[dubilbul]

Brett, I’m not in the US but I’ve browsed via anonymous proxies and got the same results. I wasn’t asking if it’s because of the update, I know it is. I just can’t understand if Google holds two different databases in one datacenter (which can explain to me many things about the Dominic and Esmeralda but will make optimizing sites more difficult), or it is querying different datacenters for different queries (this not seems to me reasonable due to the amount of traffic that the transaction will occurs additional to the synchronizing of the datacenters).

Takagi, there is no different (not even small one) between the URL’s.

[dubilbul]

OK, here is my theory. Suppose that there are only two databases (and I don’t mean datacenters) that are holding the entire indexed pages with their links-back and rank, one database for the fresh crawl and one database for the previous crawl.

The datacenters are holding only keywords and pointers to the database records and all the updating that is happen is only new sorting of the pointers based on whatever Google engineers wanted (rank formula, links back , keywords, title or what ever).

Whenever a query occur in datacenter xxx, the datacenter is retrieving from the database the information based upon the new sort of its pointers and then it projection the results for the end user (or even for itself for new sorting of keywords). Whenever a phrase is not in the datacenter keywords list, it been querying the database directly and the results are present with no sort (FIFO - First in the memory pool first out as result).

[coolasafanman]

they must somehow use a different data center for different queries. for 'buy widgets' and 'buy really cool widgets' i get a link to the index page, but each has a different version in the google cache - one about a month older than the other.

[kstprod]

I can confirm this also. All day today, I can search for keyword1 and find my page Fresh tagged at #5. I can search 2 seconds later for keyword2, and find the exact same page, with no Fresh tag. Even when I search site:www.mysite.com, both identical pages come up, one with the date and one without.

Weird