Forum Moderators: Robert Charlton & goodroi
Sometime a given datacenter may be trying out a different way to rank pages, or even dropping or adding a filter over the results. Sometimes a data center may be missing certain pages from your important backlinks -- all kinds of possibles here.
If you hover over the "cached" link, you'll see the IP address of the data center in use for you at that moment. We have a hard core group of members who watch the various datacenters regularly and post there observations for anyone interested:
[webmasterworld.com...]
I tried this search in Nashua and i got my website on page 1.
And i tried the same search in Manchester and it is nowhere.
I think and i assume that google tracks user behaviour and displays results which suits his browsing habits.
Thanks
conjo
There's still alot of theories on this and no official words as to exact facts.
Search Engine Strategies in NY had panels discussing the future of search, whiched theroized the search results would most likely show regional returns such as Google Local.
Food for thought...
[edited by: optimist at 12:50 pm (utc) on April 23, 2006]
i have noted the IP address of both datacenters using cached copy in both instances
Hi from southern NH.
I'm not convinced that looking at the IP of the cached results is a reliable way of telling which data center you're on. I don't see any reason why Google couldn't give you results from a variety of datacenters and give you cached results from another set of datacenters.
Apart from that the biggest day to day jumps have been by 10 places on a set of key search arguements. I am starting to find that while some arguements give a consistent position others seem to vary around a mean. Tests show one arguement coming out at 22, 24, 24, 25, 25, 22, 21, 23 (Ask puts it at 5 and Yahoo at 1 but that's another topic)
When you query google.com, The smoke and mirrors of DNS will route you to the nearest datacenter most of the time. So from query to query you can get results from different datacenters, with different algos, & different datasets.
So, you figure out there are many datacenters to get results from, with the possiblity of each of the datacenters reporting different results. Then you start querying datacenters and find that from a single datacenter you get different results from query to query! More smoke and mirrors, but this time it is because you do not understand what a datacenter is, and how it works.
Once again IMHO!
A Datacenter is a room full of computers (could be 10, 100, or 1000 we do not know). These datacenters are located throughout the world (remember DNS). Each of these datacenters will have a series of IP addresses facing the internet. Each of these ip addresses(referred to as datacenters), is really a single ip address on a load balancer, that directs queries to "Pools" of servers, based on the least busy server. So from query to query on a datacenter, you will get the least busy server in a pool, with the possibility of algos, & different datasets.
As gbots crawl the internet,gathering data, they return it to G somewhere, it is sliced and diced, and then pushed out to the datacenters, and distributed to the many servers there, where the "algos" decide how to rank sites for queries. Not all datacenters & servers behind the load balancers are updated at the same time. This happens all the time, hence the term "everflux"
So in an attempt to make it short(Kinda late now for that)
Ranking jumps between page 1 and page 7, possibley because the query came from different datacenters, or from different boxes(servers) within a single datacenter, many with different datasets, all in the state of everflux.
Did I make your head hurt?
Back to watching
WW_Watcher
Exactly what you described seems to happen, but the results are usually the same as when I go directly to G and enter the query.
Anyone have additional data to support the theory of queries that are not direct to G being different?