"These guys have done a lot of brilliant work on search, and ad revenue is important," Arnold says. "But neither search nor ad revenue would have been successful without the plumbing."
Google, Arnold says, doesn't store and back up data the way most large enterprises do. Every time Google collects new data, it immediately writes the information to multiple, cheap storage devices, employing what Arnold characterizes as BitTorrent-type technology. BitTorrent is common among individuals who swap files, say those containing music; corporations applying this approach to high-speed file transfers is relatively unknown, he says. "A couple of hundred million people use it during the day and find it wonderful," Arnold says.
Storage is cheap these days, and if one server fails, the data can be found on another one. Google doesn't overwrite data, it just keeps adding new information to existing files, creating very big ones. "They can serve up data very quickly."
Thus, the time-consuming process of backup isn't needed, the author says. "Google has eliminated the backup stuff that everyone pays through the nose to get," Arnold says. "They've been very clever."
Arnold says Google's storage method serves as a best practice CIOs can adopt for their companies. "This foreshadows the next generation of large computing, how companies can maximize the return of investment on their hardware by increasing data throughput," he says.
If one defines a computing era by the influential vendor of the day, the IBM Era lasted from 1962 to 1982, segueing to the Microsoft Age for the next 20 years. Since 2002, Arnold says, we've been in the Google Era.