Google Datacenters Watch: 2006-01-16

Observations, Analysis and Remarks

[reseller]

< continued from [webmasterworld.com...] >

"Celebrating" Allegra Update Anniversary Day

Good morning Folks

Life on Google DCs is much more wonderful than you will ever think :-)

Approaching February 2006, I feel like it just happened yesterday or last week or last month. I'm talking about that Allegra update which changed the map of Google serps once and for all on 2nd February 2005.

I recall loosing 75% of my Google's referrals at that black cold February day. Then the sad stories of other fellow members started to appeare on the threads.

IMO, Allegra was a turning point of the relation between the SEO's specialists and Google. Since then Google carried out an undeclared war against what most SEO's were making living of. Some wise SEO's understood at once the situation and started to adapt and deal with the new environment. Others kept dreaming about the return of the good old Google Dance days.

February is a good time for "savage" Google's updates. Better you keep your eyes opened and watch closely whats taking place on the DCs. And keep your focus on small and big changes on DCs such as

[64.233.179.99...]
[64.233.179.104...]

And lets be wise and forget all about the Smoke and Mirrors of a BigDaddy or a BigMammy :-)

I wish you all a great day and a successful week.

[edited by: tedster at 8:52 am (utc) on Jan. 16, 2006]

[edited by: lawman at 11:56 am (utc) on Jan. 16, 2006]

[Dayo_UK]

Looks like Big Daddy is appearing on and off at 216.239.37.99 and 66.102.7.99

Wonder if it will hold and spread further......

[Erku]

Does anyone know, what's happening with 64.233.179.104?

Is this going to be the future?

Thanks.

[steveb]

Dayo_UK, I'm afraid that I'm now starting to cringe when I see you say canonical problems because I don't think you are really understanding (or communicating anyway) what that means.

A canonical problem is one of the ways a page can be improperly ranked. By improperly ranked I mean that Google would want to rank the page higher if it correct understood the full algo value of the canonical page. A canonical problem is never "fixed" if a page doesn't achieve its correct ranking (unless their are other completely separate penalties or scoring demerits against that page).

Canonical problems come in all shapes and sizes. Big Duddy (sic, TM) has done nothing whatsoever to apply across the board fixes to any canonical problems ever mentioned here. Naturally some individual pages are canonically recognized properly ( or even just "better") all the time.

To speak about one thing, if the site: results show something but a page ranks absurdly wrong, then it continues to have a canonical problem. It's like the URL removal tool that doesn't remove but justs "hide" URLs. A problem is fixed when the problem is fixed: the correct URL for a page of content is ranked appropriately.

[Erku]

Easygoin,

Thank you for your help. It was very helpful. I like this new databases.

Could not find the place to leave a feedback. I think they are much better than what I have seen before.

[Eazygoin]

You're welcome Erku. If you want to leave feedback on the MC blog, click on 'comments' at the bottom, and then complete the form at the bottom :-)

[bluewidgets]

I have used this to redirect
Options +FollowSymLinks
RewriteEngine on
RewriteCond %{HTTP_HOST} ^example.com [NC]
RewriteRule ^(.*)$ http://www.example.com/$1 [L,R=301]
and is working if I type [mysite.com...] I get www.mysite.com
Google also has picked up the 301

what is the diference if you put
Options +FollowSymLinks
RewriteEngine on
RewriteCond %{HTTP_HOST} ^example\.com [NC]
RewriteRule ^(.*)$ [yoursite.com...] [L,R=301]

[rainborick]

The only difference is the escaped dot ("\.") in the URL in the rewriteCond. The syntax for the .htaccess file often baffles me, but it seems clear that the intent of both versions is identical and should be functionally equivilent as well, providing the syntax is permissible.

[ScottD]

Dayo_UK

216.239.37.99 gives me the same result for my "problem" search as Big Daddy too, so hopefully its spreading.

[colin_h]

Steveb is right & Google still can't get it right and put me at the top of every listing ;-)

Those pesky Canonicals!

Less Yap, More Nap ... Goodnight All

[phpdude]

Sounds to me like Dayo_UK us going by Matt's description of what a canonical problem is:

Q: What is a canonical url? Do you have to use such a weird word, anyway?
A: Sorry that it’s a strange word; that’s what we call it around Google. Canonicalization is the process of picking the best url when there are several choices, and it usually refers to home pages. For example, most people would consider these the same urls:

www.example.com
example.com/
www.example.com/index.html
example.com/home.asp
But technically all of these urls are different. A web server could return completely different content for all the urls above. When Google “canonicalizes” a url, we try to pick the url that seems like the best representative from that set.

In that example, it is made to look like it is the [mysite...] verses the non [mysite...] indexing issue.

It's easy to see why people can interpret it differently.

[steveb]

Matt's example does explain the most common canonical problem. But don't make it more mysterious than it is. Google already always canonicalizes everything. That's part of what they do. The point is them canonicalizing CORRECTLY.

For example, Google might already see www.example.com as the canonical page, but if it isn't recognizing a 301 from example.com and www.example.com/index.html to www.example.com, then a canonical problem exists. Getting the page right is only prt of it. If the correct page should score 537 but instead scores 142, that is an ongoing problem, regardless of whether Google's current canonicalization has chosen the right page as the canonical page.

Identify the right page. Rank the righ page correctly. That is what Google does the big majority of the time successfully. Unfortunately, they do not do this sometimes.

(In other words, choosing the wrong canonical page is just one of several possible canonical errors.)

[afterburner]

I am trying to turn off my email notification of replies for this thread but I cannot find the owner edit button to do it, it is not on the left beside my profile? Is there another way to back out of all these emails? Thanks

[BillyS]

While I believe that steveb is technically correct, I do not think this is necessarily how most webmasters have come to use the term.

Canonicalization is the process of making something so that it conforms to all the rules or specifications that it should conform to. I think when MC is speaking this way, he is referring to the ability of Google to recognize a certain page for what it truly represents.

Once that is accomplished, then all the rest should flow from there. I think this is the point steveb is making. If Google can get the first part right, then ranking a page should naturally follow. But that has to happen or we've just got a patch on a much larger problem. In other words, Google has the ability to "pretend" it has fixed the problem by manipulating the way pages show via site: commands. But just fixing that command does not really solve the problem.

[steveb]

One example. I'm looking at a site where www.example.com is the first page listed for a site command. The second page listed is a supplemental of example.com. Prior to Feb 2005, the two pages were canonicalized as www.example.com and it ranked #21 for a hyper competitive term. When this index page was essentially "de-canonicalized" in Feb 2005, and the www and non-www were split into two listings, the page dropped hundreds of spots in the results.

Listing or identifying the www first is of no consequence. Canonicalizing the two listings (including obeying the 301 from the non-www to the www) includes scoring the page properly. That is after all what matters... the search engine ranking pages in a user-friendly way.

[ramachandra]

For my site I have written this code for 301 redirect

<%
PathInfo = Request.ServerVariables("PATH_INFO")
ServerName = Request.ServerVariables("SERVER_NAME")
Iswww = InStr(ServerName,"www.mysite.com")

If Iswww < 1 Then
NewLocation = "http://www.mysite.com" & PathInfo
Response.Status="301 Moved Permanently"
Response.AddHeader "Location", NewLocation
End If
%>

In this code I think there is a problem, the code is permanently redirecting to www.mysite.com/index.asp by this line in the above code (NewLocation = "http://www.mysite.com" & PathInfo), & pathinfo is storing the index.asp and adding it to the header, I am thinking this might be the reason which cause canonicalization.

Now I have removed the "& pathinfo" from the code.

Am I thinking right?

[reseller]

Good morning Folks

Very interesting rich educating canonical discussion!

Just wish to say a BIG THANK YOU for your contributions. And of course special thanks to my kind fellow member steveb for his high spirit of sharing!

Wish you all a great day and God bless.

[heartless09]

I have been reading a lot in this forum. I am a new webmaster trying to get indexed in google.
I have made a css pure website. Validated for css and xhtml. I have made a sitemap. First a text file and a few weeks later a xml file.

Finally google indexed my website a few day ago. First it showed only the index page but yesterday for a shot time it showed more pages of my website.

I checked it today and it shows only the index page. The datacenters 66.102.7.99 and 216.239.37.99 are showing more pages.
Does this mean something?

It is a fresh new domain.

[Dayo_UK]

Steveb

Yes, I understand - which is why I was talking about Homepage Canonical Problems.

>>>>>>For example, Google might already see www.example.com as the canonical page, but if it isn't recognizing a 301 from example.com and www.example.com/index.html to www.example.com, then a canonical problem exists.

For me this is what is now happening for the homepage G is recognizing 301s when they are in place, and even when they are not - but the rankings are not returning.

As MC says it usually refers to homepages and a large number of sites have this problem just for the homepage

>>>>>Identify the right page. Rank the right page correctly.

This has been what I have been saying - to me identification of the correct page (or so it seems) is an improvement - perhaps I am being to optomistic that rankings will follow naturally in time.

>>>>>If Google can get the first part right, then ranking a page should naturally follow. But that has to happen or we've just got a patch on a much larger problem.

Exactly - as I have been saying.

MC says these are subtle changes and rankings changes are not likely in this interation - or along those lines.

Say the ranking changes are based on internal PR or another scoring method that has not been added to the ingredients yet on the BD dc until they are happy with the identification process.

For me Canonical has always been about identifying the correct page - rankings are related but seperate which is why I say I am see Canonical improvements. But I did not say that the problem was fixed - as I have said a few times the ranking has not returned/improved.

So MC/GG - What the are we supposed to be looking for at this stage - please clarify ;)? Improvements in the identification process or improvements in ranking as a result of Canonicalization changes?

Hmmz - BTW BD did not spread last night - it did hit a couple of dcs for a while but this has not held by the looks of it.

[ScottD]

heartless09

If Google has only just visited your site, it probably won't be showing all your pages yet, and the amount it shows may fluctuate with different datacenters. With the next visit that should improve, but showing up in serps may be 6 months away, or even more. Try to concentrate on building the site for now, and forget about Google. Be glad it has found you. The rest will require a lot of patience.

Big Daddy

You should get out more. Why not take up salsa classes or something? Theres a whole world out there waiting to meet you.

[Miop]

<For me this is what is now happening for the homepage G is recognizing 301s when they are in place, and even when they are not - but the rankings are not returning.>

I see two things happening to my site -

search [mysite.com...] shows that non-www url home page but with site pages all showing correct www. Site ranks poorly.
Where g is no longer showing that URL at but automatically shows www version, rankings are much better.

The latter is the case on only one dc - 64.233.179.104

On 66.249.93.104 non www is still showing for home page though rankings are better than any other DC except the other one.

I thought G had recognised the 301 correctly when it began showing all www pages for the non-www search, but that is not the case.

I think I am now having a similar case with the index.php/ root page issue - rankings are great where the 301 has taken effect correctly, but not where it hasn't.

[reseller]

Not All BigDaddy DCs Are Created Equal!

Hi Folks

I'm sure some of you might have noticed the same. At present there are 2 sets (4 DCs) of BigDaddy DCs which are showing stable results. However, the serps of the 2 sets don't look the same, as far as my testing keywords are concerned.

Set-I
[66.249.93.99...]
[66.249.93.104...]

Set-II (reseller's DCs :-))
[64.233.179.99...]
[64.233.179.104...]

Do you see any difference between the two sets too?

Hi Matt "Inigo" Cutts and GoogleGuy!
Any weather report? Thanks.

[Miop]

yes - the latter set is no longer showing non-www (in my case) correctly.

[vdoyl]

Yep, two sets they are :)

Earlier today, I think they were even 3 sets.
At least site:domain.com was showing 3 sets of results for me.

Now, I see there are only 2 sets as described by reseller.

They are shifting the data for a couple of weeks N-times a day - I wonder when they will have a kind of stable index :)

[colin_h]

Hi Reseller,

The two sets are behaving as I suggested might happen on [webmasterworld.com...] .

All of my supplimentals have gone and replaced with a lovely fresh set of cache.

Brilliant, I love it when I guess and hit the spot.

[cleanup]

Set "I" have the most upto date results and contain the 301 rubbush for my pages.

Set "II" have the pre-jagger results for my site.

[paintbox]

For me site:domain.com shows non-www for the home page and www for the rest on set I. On set II all pages are www. Serps look the same. I have no redirect in place.

[g1smd]

If I keep hitting refresh (waiting a minute or two each time) on http://66.249.93.104/ then I see three different results (different number of results) returned.

[reseller]

g1smd

>>If I keep hitting refresh (waiting a minute or two each time) on [66.249.93.104...] then I see three different results (different number of results) returned.<<

Do you also see three different top 10 sites sets, or you see the same top 10 sites when hitting refresh?

Thanks.

[BillyS]

>>> [66.249.93.104...] then I see three different results (different number of results) returned.

I see that too. On both sets, my page count is starting to increase up to 585 from 485 (shows 985 on non-BD DCs).

[Dayo_UK]

If I keep hitting refresh then very occassionally I dont get Big Brother results on that DC - as in same results as the other dcs.

Cant see a third set yet - Although I agree there is a slight difference between the BD results.