|
APRIL 2004
Google's Search Engine, Google's
Change, Google's Data Center, Google Facts
Google uses something called distributed architecture (distributed
computing). Google links over 15,000 normal pc's together
across the globe via approx. 12 data centers. Our theory is
that the network is broken down into 3 main areas, the core
layer, distribution layer and the access layer.
At the core layer is the network back bone, this links up
data centers via the distribution layer. The access layer
is offered to the search engine interface (what you see when
you visit google) for traffic direction and searching traffic.
The core layer will probably connect to the global Internet
via high speed connections.
The importance of such a design can add extra layers for
redundancy, overflow routing for high traffic peaks, and the
15,000+ pcs spread out over the 12 or so data centers can
be synchronised across this design on a regular basis.
The Google Dance
In the past each data center would go through a massive update
each month. When the update was in a 'during' phase people
could see major shifts in the SERPS (Search engine rankings).
This was called the Google dance.
Further Explanation
When you type in Google into your browser, your pc sends
a DNS
(i.e. www.google.com) request to a DNS
server (configured within your IP
stack). Most ISP's offer something called DHCP (Dynamic
Host Configuration Protocol) - this will set up your default
gateway route and will tell your pc what server to send
the DNS request to.
The DNS server resolves the domain name (translates the URL
to an IP address). IP or IP addressing is more than just a
number, at the hart of the Internet is something called an
IP
stack. IP protocols (most networks use version 4) will
encapsulate your data into layers of organised information
cells or packets for streaming. For instance destination address,
source address, and other identification information (bit
settings) such as TTL. (Time to Live)
When you do a trace route on the network the application
sets the TTL to 1 and this counts down on each hop. When it
gets to the first hop the TTL expires and ICMP sends back
the response to your host. The application then sets the TTL
bit to 2, then 3 and so on. Eventually you see a map of the
network. If you go to to DOS and type in tracert www.google.com
you will see each hop that connects you to the access layer
device on Google. When your browser sends a request to www.google.com
the TTL field (within the IP header) is set to 5 minutes,
and the request is handled by the closest geographically placed
server. If the server is down (i.e. hard down - has a network
complete failure or just down - has too many requests) google
redirects your request to another data center.
WHY
NOT WORK WITH US TO BUILD A BETTER INTERNET - OUTSOURCE YOUR
SEO WORK TO US
When each data center is at a different phase of the synchronisation
process it is possible to see changes within the search results
i.e. you search for cars on google and see a site at number
3, you then click refresh and the same site is at number 6
etc. Some of these ranking shifts could be quite dramatic,
as a result Google now make smaller adjustments more frequently
to control this problem.
The dramtic changes before were nicknamed the Google
dance, as positions would dance around the rankings
table.
When Google visits your web site, it makes an impression
of each web site it stores in a data center, this impression
is known as a cache. When you switch between data centers
on Google you access information from different caches. The
Google Dance is no longer alive in the sense that Google do
not do a major update of cache synchronisation every month,
they simply do smaller updates more frequently to keep the
information up to date and accurate.
This process in real-terms is a Google convergence process.
Current List Of Data Centers
| www-mc.google.com |
66.102.7.100 |
| www-lm.google.com |
66.102.9.100 (new) |
| www-kr.google.com |
66.102.11.100 |
| www-ex.google.com |
216.239.33.100 |
| www-sj.google.com |
216.239.35.100 |
| www-va.google.com |
216.239.37.100 |
| www-dc.google.com |
216.239.39.100 |
| www-fi.google.com |
216.239.41.100 |
| www-ab.google.com |
216.239.51.100 |
| www-in.google.com |
216.239.53.100 |
| www-zu.google.com |
216.239.55.100 |
| www-cw.google.com |
216.239.57.100 |
| www-gv.google.com |
216.239.59.100 |
Current List Of Test Centers
If you want to get an idea of where you site is going to
be placed a good trick is to visit www2.google.com
or www3.google.com
They will show you a more accurate picture of what you can
expect so see in the real SERPS.
Google Tool Bar
Google offer something called a browser tool bar that is
a great utility to see what google think of your site, this
can be downloaded here: http://toolbar.google.com/
Your tool bar connects to a data center, if you want to redirect
this to a particular data center simply add a comment to your
host file (or lmhost file) as follows:
216.239.59.100 toolbarqueries.google.com
Obviously you put the IP address of the data center you want
to connect to.
Google
Changes
Google's Technology Playground
Google's
Page Ranking Explained
Website
Business Reports Here!
Web Site Ranking Explained,
Web
Site Submission, Search
Engine Optimisation Basics,
Internet
Advertising, Google Search Engine Optimisation
Guidelines,
Flash
Intros, Web
site Templates, Search
Engine Relationship Chart
Search engine marketing is
needed to successfully place your site on sites such as Google,
Yahoo etc. unless you use PPC. If you are a long term PPC
client the SEO will be better for your business long term.
|