Forum Moderators: Robert Charlton & goodroi
Any IP addresses would be most welcome.
So... you cannot see the future on any ip address, and ip addresses do not correspond to data centers at any rate.
Matt Cutts once said that for most of the time, all the IPs in one Class-C block are usually mostly in step with each other, but under heavier loads, faults, downtimes, upgrades, or other random stuff that isn't always true.
From time to time Google picks one or two class-C blocks to run some experiment on. Sometimes different experiments on different blocks at the same time. On some occasions those have been a window to the basis of results that spread to the rest of Google some weeks later. Several times they have been very different results, but then those have not spread.
Most times that Google has made these experiments, they have been on different Class-B blocks each time. On occasions the results changes may have been so subtle that many people may have missed the fact that any sort of experiment or change was going on at all.
At the moment, Google is updating their Directory. The new version is only seen on a few Class-C blocks. Over the next few weeks I expect that it will spread to most of the others (there are a few that do not have any copy of the Directory at all, as opposed to only having an old copy).
[edited by: g1smd at 12:25 am (utc) on Aug. 23, 2007]
For example : 64.233.171.99 is one DC. Found a list here too: < url removed >
My question is which ones change most frequently regarding SERPs.
Some have to be more active then the others... and some people here must have noticed them?!
[edited by: tedster at 1:10 am (utc) on Aug. 23, 2007]
If anything, then 64.233.171.xx might be a "datacentre"; or one "block" or "floor", or "room" of machines.
There are many values for xx that are valid for each block. Google uses at least 44 such blocks.
All the machines within any Class-C block of IP addresses serve the same results as each other for the vast majority of the time.
Machines from some other Class-C block may display the same, slightly different, or very different results at that exact same time.
Just ran a test < with an online tool > and had different results on some DCs.
I assumed some of these DCs are known to take the lead when changes occur and some of you might know them. But as you said those that take the lead are chosen randomly so no established DC leaders are known?:)
[edited by: tedster at 1:08 am (utc) on Aug. 23, 2007]