PARAGOOGLEZATION A Brief Look On How Google Utilized
PARAGOOGLEZATION A Brief Look On How Google Utilized Hardware Over Time
� Two Ph. D. students from Stanford University, Larry Page and Sergey Brin, started working on an web search engine project in 1996; named “Backrub”. � In September 1997, google. com domain was registered and Google project was launched at a garage. � The original hardware used by Google in 1998 was composed of: • Sun Ultra II: dual 200 MHz processors, 256 MB of RAM. • 2 x 300 MHz dual Pentium II servers, with 512 MB of RAM. 9 HD’s with 9 GB capacity were shared between these two servers. • F 50 IBM RS/6000: 4 processors, 512 MB of RAM, 8 x 9 GB HDs. • Various disk expansion boxes which were consisting of: 3 x 9, 6 x 4, 8 x 9, 10 x 9 GB HDs. History
The First Server of Google in 1999 Each level of this rack has networked computers with cheap hardware; some of which were partially overlapped and caused damage. History
� It is rumored that today, over 450. 000 servers are used by Google in many major server farms around the Earth. Each of them are made up of low-cost computers that are running a modified version of Linux, which is actually based on Red Hat distribution. Server farm: A collection of servers which are maintained by an enterprise, commonly used for cluster computing. They are increasingly being used in addition to mainframes. Today
� Today, Google use their own distributed file system (named Google File System, or GFS) on their server clusters. � A distributed file system is a network file system in which a file can be accessed from only specific nodes on the network, in contrast to shared file system where the whole network may access the file. � GFS ensures that at least three copies of a file is stored on different computers of a cluster. � If one of these machines fail to access a file in milliseconds when requested, the cluster immediately asks the other computers to retrieve this file. Today
� The exact configuration of the network and the location of all the clusters are unknown. � Some facts shows that the network is composed of 450. 000 servers with: ◦ Processor: As of 2005, processors ranging from 533 MHz Intel Celeron’s to dual 1. 4 Intel Pentium III. ◦ Today, to reduce costs and increase stability, processors are switched to AMD Opteron from Intel Xeon; but this is not confirmed officially. ◦ Disk Storage: At least two hard drives with more than 120 GB storage per computer. ◦ Main Memory: More than 4 GB of RAM per computer. Speculations
� Google is currently developing a supercomputer codenamed “Project 02” at a datacenter in Oregon. � Many details about Project 02 are kept secret today, but it is expected to be used in the network which processes billions of search queries per day and many other services provided by Google; thus will increase the effectivess of these products. Future
- Slides: 7