The amount of data processed by Google on a day is 20 petabytes or more

According to the announcement by Google Fellow, as of September 2007, Google seems to handle data of over 20 petabytes (20,000 terabytes = 20,000,000 gigabytes) a day on a large computer group. Even so, it's hard enough to imagine the amount of data.

Details are as below.
Processing number and data comparison in August 2004, March 2006, September 2007. I understand that the number of jobs (Number of jobs) and the number of machines (Machine years used) have increased considerably, and the processing time is decreasing conversely. Also, the amount of data has increased dramatically.


In addition, the specification of the machine used by Google is "2 GHz Intel Xeon processors with Hyper-Threading enabled" × 2, 4 GB memory, "160 GB IDE hard disk drive" × 2, Gigabit Ethernet. This type of machine cost per machine is $ 2,400 (about 260,000 yen), and the monthly hosting cost is 900 dollars (about 100,000 yen). And it seems that cost of just 1 million dollars (about 100 million yen) is required even for hardware cost alone.

Google processes over 20 petabytes of data per day