I wouldn't worry about getting flagged for duplicate content, since googlebot will only go to one site, depending on how up to date its DNS is.
It might, however, be a good idea to mirror new content on your old sever until you shut it down.
If I were to switch, I'd do it at least 2 weeks before an aticipated deep crawl and keep the old server up for 2 months, although one month could be enough.
Thank you for your reply 4serendipity,
I take it then that once Google finds the site on the new server, it will not go back to the old IP address.
I know that the Google update might take place any day now. Am I correct to think that changing the DNS during an update has no adverse effects in indexing and ranking?
|... since Google and other bots will keep visiting the old IP address until they update their IP number databases ... |
That's interesting cyberfish. Google visits you by IP numbers? and not the canonical name? I have just moved a site to a new host (as of friday) and have left duplicate content up at both places.
Today I noticed that Googlebot is still going by the old site while 90% of the traffic has shifted to the new host server. I guess I'll just watch the logs and see when Googlebot and all the other SE bots switch over to the new host and then I'll terminate the old hosting Co.
Thanks for posting this. Hopefully someone with more experience with this will comment.
I haven't heard of this before. Many times a webmaster does not have a choice about leaving their old site up for extended periods.
Does this also apply to Yahoo? Thanks.
>That's interesting cyberfish. Google visits you by IP numbers? and not the canonical name?<
I was surprised as well to find out that Google uses a cache of IPs which seems to be updated every month or so. I am not sure exactly how Google uses it to be honest. I understood that it is more efficient for the googlebots to use this cached IPs.
The Google Knowledge base has a short blurb on it:
Doing a search for host moving or server moving yielded a lot of talk about it here at WW.
I am still cloudy on the matter, and although I have uploaded my site on both servers, I did not change the DNS yet as I am trying to understand the ramifications.
Two more threads that might be of interest:
If Google visits by IP address wouldn't that mean any virtual sites would never be indexed?
And the newest hyper modern advice - make sure the old ip goes 404 or 500 asap. Evidently, it gets gbot into a new mode and replaces the old ip with the new in the da master list.
Kryton - some virtual servers offer you a unique IP for your domain
For what it's worth...
I moved hosts about a month back, Googlebot was virtually the only bot etc. to visit the old site once the DNS had resolved to the new host.
Last week the freshbot stopped visiting the old site and found the new. I have no idea if the deep crawl will be of the new site or the old but I have both sites up so it's no big deal.
Google is slow to update its DNS cache so you have to wait for it to find your new site/host or follow Brett's advice and serve up a 404 ;).
While it is not visible in the normal user view of a browser request or much discussed, all HTTP requests are accompanied by a Host name field which allows virtual hosts sharing an IP address to be resolved by the hosting server.
For more details, see RFC2616, page 36 at [ietf.org...]
The normal request sequence is that a browser requests a domain name translation (if required), the DNS system returns an IP address, and the browser then requests a resource from that IP address, including the Host name field in the request.
So yes, all resource (page, script, image) requests are made to an IP address, not a domain name. Google caches the IP address associated with the domain names it crawls in order to speed up crawling and minimize the otherwise enormous load it would places on the DNS system.
ok thanks jdMorgan.
I have transfered servers a number of times, the last time I kept my old server for only 3 days. But I was in control of the DNS servers my domains pointed to.
cyberfish, what I did was keep the old server until I could see googlebot visiting my new server in the logs (I have a few PR6 sites which are crawled daily, and I noticed google changed to crawl the new server/IP after one day). First I copied my old sites to my new server. Then I update the DNS servers to point to the new server. I waited until the DNS propogated, and googlebot visited the new server/IP. Also you could keep your log analysis software running on both machines, and confirm you are getting roughly the same amounts of daily visits on your new box as you were on your old box.