Carbon-Pros Analyst Blog

Scalability allows a system to be gracefully upgraded to process more transactions by adding new processors, devices, and storage. Google is the king of scalability. It indexes billions of web pages to support its search engine. It operates hundreds of thousands of servers worldwide and processes millions of search requests each day.

Google refers to its architecture as a three-layer stack, most of which was developed in-house. At the top of the stack are Google software services such as search, advertising, email, maps, and many others. In the middle of the stack is their distributed system architecture. This includes the Google File System (GFS), a distributed storage system (Bigtable), and a programming model (MapReduce) to support parallel processing.

At the bottom of the stack are the hardware and OS. Google operates hundreds of thousands of machines in 30-40 data centers. Each machine uses a standard AMD/Intel x86 chipset running a customized version of Linux. Each server has its own 12-volt battery to supply power in case the main power supply goes down.

Google does not say how many servers they operate but observers have estimated that the power required for a half-million servers ranges upwards of 20 megawatts and would cost in the neighborhood of $2 million per month in electricity charges. Google has an obsessive focus on energy efficiency. Fortunately, they have recently started to share some of their knowledge with the rest of the world.

If you wonder what this has to do with smart grid, it's simple. The smart grid needs to be architected for massive scalability with millions of devices including meters that transmit readings as often as every 15 minutes. This particular note will appear in the book as a sidebar relating to our discussion of enterprise application architectures.

Sources: High Scalability, Wikipedia, and CNET
http://highscalability.com/google-architecture
http://en.wikipedia.org/wiki/Google_platform
http://news.cnet.com/8301-1001_3-10209580-92.html