Forum Moderators: Robert Charlton & goodroi
Examples:
[webmasterworld.com...]
[webmasterworld.com...]
I apologize for length, but I think some clarification about the BigtDaddy issue is needed.
BigDaddy is more than an update.
It's the adoption of a new "infrastructure" plus (probably) some changed software. While nobody will tell us what they mean by "infrastructure", my guess is that by "new infrastructure" they mean the migration of servers to their own version of Linux (Goobuntu). Another guess is that the new linux has some custom or otimized features that alow to do other things with the Google software (things like running their algos more deeply to better filter spam etc.)
Being something more radical than a simple upgrade, it will not spread as fast as usual to all the data centers.
The “BigDaddy infrastructure” has been running since the end of november on a single "test" data center; at the start of january they have started implementing it on "live" data centers. New data centers are being migrated at the rate of about 1o days each. Assuming a total of about 20 data centers, it will take a good part of the year to cover all of them.
How do Google datacenters work?
Usually, when you type www.example.com a DNS lookup takes place: the domain is searched, its corresponding IP address is found (eg. 123.123.123.123) and the request is sent to server 123.123.123.123, hosting www.example.com
When you type "www.google.com" something quite more complex happens. Depending on a few factors you are redirected to a different data center. The main factors are geographical location (you tend to be served the centers that are physically near) and load (if a data center is overloaded, you are switched to another center)
So, whenever you go to www.google.com you can see an "old" data center or a "BigDaddy" data center and its' not easy to tell them apart.
But IN THE AVERAGE, as your users are seeing a random mix of data centers, you should see a gradual introduction of the Bigdaddy effect. For example, we can assume that on jan 4th the first Bigdaddy center went live and so approximately 5% of the global Google traffic went through BigDaddy. When 10 data centers will be under BigDaddy, 50% of the traffic will be through BigDaddy.
So, if since the beginning of january you see a gradual trend in change (be it a good trend or a bad trend), that's probably the Bigdaddy effect, and (again my guess) expect it continue in the same direction for a few more months.
If on the other hand you see a sudden change on a specific day, that's probalby something different, the usual Adsense rollercoster.
A note: if you are in a non-US country, then (presumably) when the nearest data center will migrate that will generate a strong, sudden, variation.
Also, bear in mind that BigDaddy doen not related to Adsense specifically, as far as we know. I only effects the Google searches, and all it can change is traffic. If traffic is the same but your earnings change don't blame BigDaddy!
I'm in the UK, in January my adsense doubled, I put it down to a freak peak but as it is my first year I don't kbow whether it is post Christmas or could it be the BigDaddy effect?
Who gave it that name? The only BigDaddy I know of was the late Shirley Crabtree.
Shirly Crabtree? He was a big, big wrestler from the 70's
Yay, he actually started in the 60's through to the 80's, literally a big, big wrestler, not quite as big as Giant Haystacks though!
A guy (JeffM) there suggested BigDaddy because that's how his kids call him.
Jafo, the BigDaddy info mostly comes from Matt Cutts' blog, a must-read for the latest Google thingies.
Also, Forum 30 here is constantly speaking about it...
The Goobuntu bit si around the 'net since a few days, with some guys hoping Googel would actually distribute (which is unlikely, by the way and if you want it get the original Ubuntu, unless you need an OS tailored for ditributed processino on networks of hundreds of PCs)
The link between thew two is my guess, which might be wrong of cource...
The link between thew two is my guess, which might be wrong of cource...
still great thinking
and complements on giving best explanation about BigDaddy
It's the adoption of a new "infrastructure" plus (probably) some changed software. While nobody will tell us what they mean by "infrastructure", my guess is that by "new infrastructure" they mean the migration of servers to their own version of Linux (Goobuntu). Another guess is that the new linux has some custom or otimized features that alow to do other things with the Google software (things like running their algos more deeply to better filter spam etc.)
There are at least 3 types of machines. Crawlers, indexers, and data servers. In Googles system you can also add machines for page rank calculations and that is done separately from the indexing.
Data servers are what we call “data centers”. The infrastructure, “software”, of a data center is an algorithm that determines and returns results relevant to the query.
Ranking factors are applied to the entire database during indexing and other processes such as page rank. However, the final bits of ranking and relevance can only be applied on a data center, after it receives a query. It is this final piece of software that determines which site will be number one from a query subset.
Google has been running a modified Linux operating system for most of their existence
Yes, but I think the partnership with Ubuntu is the new factor here.
The infrastructure, “software”, of a data center ...
I think that reading "infrastructure" as software is limited.
If it were only a "normal" algo change, Matt Cutts would have called it an update, as always.
I get the idea that here we are speaking of something underlying the normal google software...
This is what he says:
Q: Why are you giving this a name? Isn’t that normally the privilege of Brett Tabke and the moderators at WebmasterWorld (WebmasterWorld)?
A: Brett and WebmasterWorld normally name updates. But this is neither an update nor a data refresh; this is new infrastructure. It should be much more subtle/gentle than an update.
Also, in no other case it took 10 days per data center to spread an update of the google software. Again, quoting Matt Cutts:
I’d expect a new data center to be converted to Bigdaddy roughly every 10 days or so.
The picture I get is the one of a team of guys that get to a data center, start installing the new OS on each machine and move to the next one ...until the last one is converted.
But again, I might be wrong...
You mean Haystacks Calhoun, who was noted for the "big splash." Perhaps that's where BigDaddy came from.
I wish I knew for sure that this was a result of Big Daddy. If the increase in traffic that I'm seeing is a result of Big Daddy and BD results are currently only on 3 DC's, I can hardly imagine what it will be like in mid-March when Big Daddy is used at all DC's.
I admit, it's hard not to put all my eggs in one basket when the eggs are from a goose which is laying golden eggs. I guess time will tell.
In business, you must have your employees operating within a standard system. All distributions of Linux are extendable, and Google has done that to better handle its data. Goobuntu, I would think, would be more to do with a standardized desktop interface for Google employees to communicate to each other and run the business of Google. Regardless, this news is good news indeed for the Ubuntu and Debian communities. You can’t buy this type of endorsement.
Ubuntu offers a live CD version if you are curious enough to try it out. You can download, burn it to a CD and run it from the CD.
As your post and observation suggested the spread will take time, so January traffic change to few of us might just a generic change not BigDaddy.
I'm not seeing any real difference in my rankings on the many keywords I've checked lately yet my traffic is up. Could there just be more people surfing the web?
The fact that the changes are in the "infrastructure" (be it OS or anything else) does not mean that software running on the infrastructure will not change, i.e. that the SERPs will stay the same.
Matt Cutts, for example, gives the following examples about what will change in the SERPS:
... the new infrastructure at the Bigdaddy data center will let us tackle canonicalization, dupes, and redirects in a much better way going forward compared to the current Google infrastructure.
I find the reference to "dupes" particularly interesting. first of all because already there are reports of SERPs shaking for duplicated contents, and second because the detection of duplicate contents is a typical example of an operation that might (MIGHT) require a stronger "infrastructure" (be it OS or whatever) under the software so that it can run "deeper", for example with more iterations.
Another interesting but cryptic hint is the way Matt Cutts goes on:
I’m not claiming that everything is perfect in Bigdaddy, just that it’s easier for us to make changes and improve search quality as we get feedback from you.
The news of the banning of bmw.de from google is of otday, but already since january 11th Matt Cutts was speaking of automobile.de, and said "In 2006, I expect Google to pay a lot more attention to spam in other language".
My intepretation of this phrase is: "In 2006, the new "infrastructure" will allow Google to pay a lot more attention to spam in other language, devoting computing power to these "minor" languages too".
BTW, This capacity makes me giggle with joy, because the italian SERPs are quite often manipulated by low-level SEO techniques, and my white-hat efforts were up to now quite easily "crushed" by extra-spammy pages!
