This is a phase in which "cloud computing" replaces independent computers.

Absrtact: If you combine the Internet with cloud computing, you'll have a complete machine that completes three important tasks in the brain: information storage, processing, and communication I don't know what the Internet will be like in 50 years. But I do know it for the next 10 years to 20.

If you combine the Internet with cloud computing, you will have a complete machine that completes three important tasks in the brain: information storage, processing, and communication

I do not know the development of the Internet in 50 years. But I do know what it's going to be in the next 10 to 20, and I'm sure its development will be based on the theory of brain science. Other competitors in the industry foresaw this fact.

It is no coincidence that Google's Larry Page and Sergey studied artificial intelligence at Stanford University, and that he was an outstanding authority in the field from Trie Winograd, and that they had put their studies into practice and formed and developed their company. (In fact, Google's 18th employee is a brain surgeon who later became Google's operating director.) )

In fact, the most successful internet companies are directly or indirectly premised on the belief that the Internet will develop more and more like the human brain.

The next round of change in the computer begins, a phase in which "cloud computing" replaces an independent computer. Millions of companies are choosing to use renewable "cloud" resources, and the old parts of the cloud will soon be replaced by newer and faster parts, and "clouds" are like life-moving creatures.

Google's Secret complex

A city called Dallas, where people found a mysterious building standing in this barren landscape. The building, which has two rugby fields, stands over 4 cooling towers, surrounded by high barbed wire fences.

When the construction started, Dallas residents only knew that the building was built by internet search giant Google, and the rest was unknown. After learning that the IT manager of the nearby OVF company (Orchard View Farms) and several other locals joined, there were a number of guesses. When the new Google employees who signed the confidentiality agreement were unwilling to describe what they saw and what they were doing, the speculation was stronger.

Google finally decided to invite reporters and editors from the Dallas Chronicle (Dalles Chronicle) to interview. If these journalists really go deep into the core of the building, they may see that thousands of computers, connected in metal boxes, have 40 or 50 central processing units floating in large quantities above the air and up to 4 cooling towers. They may also see gray cables and blue, orange Ethernet cables winding across the building, connecting personal computers, with about 100 IT staff serving the needs of these PCs and wearing soundproof earmuffs to protect their ears from the noise of fans.

If you think about the scenes in the movie "Matrix", where many people are gathered to collect electricity, then your imagination is correct, but those people are replaced by computers.

The internet will become larger and bigger

Why is it so mysterious? Part of the reason is the competitive advantage of the consideration. By connecting thousands of ordinary computers, Google builds a supercomputer.

The Dallas server cluster (also known as a multiprocessor cluster) is monetize and possibly the world's most advanced achievement in parallel Computing: using many computers to accomplish multiple tasks at the same time. The Dallas server cluster appears to be Google's most advanced achievement, but it is not unique: Google has 24 large server clusters across the United States. Even if the details are not known, the total number of servers is likely to exceed 500,000, with 200PB (petajoules byte) hard disk storage capacity and 4PB random storage capacity. Think about it, 1PB equals 10,000,004 bytes, and your ipod can hold 100多万首 songs.

By using a vast system of multiple server clusters to carry out a vast computational effort, Google has created "cloud computing", where thousands of of inexpensive computers gain huge computational power by massively parallel processing. The point is that it's not only cheap, but it's faster to get information from random storage on a regular computer than to get information from a supercomputer-intensive disk drive.

Google's "cloud computing" began in Kristov Bisilia, a 27-year-old computer engineer. Google has developed a distributed computing system (the core technology of cloud computing), a proprietary software that splits a task into countless subtasks, executes it by a large number of computers, and finishes processing in one-zero seconds. Now more than Celia, Google's president, Eric Schmidt, has come up with a new idea: Build a mainframe computer from a small personal computer using the basic functionality of distributed computing systems. His idea was encouraged by Schmidt.

Then, IBM's president, Sam Pamisano, also gave support to--40 computers connected-the experiment began. This year is 2006 years. Now Google has 500,000 computers working in parallel. In fact, companies (including Google, IBM, Microsoft, Yahoo, and Amazon), which are building clusters of cloud computing servers independently, now have more than 2 million computers working at the same time.

These machines consume as much electricity as Las Vegas in hot summer days, so the challenge for these companies is to find more (cheap) power resources. Coincidence, China is considering building a series of nuclear power plants, and the companies that provide cloud computing are already impatient-though they are reluctant to say so.

Where is its future? Mark Dien, IBM's research director, once said that with the development of cloud computing capabilities, the Internet will become larger. "The Internet is too small," he said, "and one day we will laugh at its insignificance." ”

The intelligent development direction of Internet

Google and other cloud computing service providers want their cloud computing to have some kind of human intelligence. In other words, they seek better distributed computing, a goal that is consistent with the pursuit of brain scientist Jim Anderson and his artificial brain team. The goal is to create a circuitous, repetitive pattern of thinking in the human brain.

Google's work begins with distributed computing systems, a simple, powerful software program that can automate parallel processing and large-scale computing work. In other words, it allows Google to use "cloud" resources and the Internet to think in parallel. Distributed computing systems work the way our brains do: classify important information, then distribute it into a service group of personal computers, delete irrelevant information, and computers--unlike distributed computing systems and the brain--Receive all the information.

Distributed computing systems have abandoned the mechanical nature of traditional computers (as Google's senior vice-president recently boasted, "We are building a server cluster that is more flexible than any other company"). Google Now applies distributed computing systems to tens of thousands of of programs, from satellite image processing, large-scale machine learning issues, to language processing and answers to frequently asked questions. It handles approximately 100,000 different functional applications per day and analyzes 20PB bytes of data.

The inventor of the distributed computing system said in the latest article: "It has been widely used in various areas within Google, including: Large-scale machine learning problems, clustering problems, extracting data for common problems of the report, extraction of the characteristics of Web pages for new experiments and products; satellite image data processing Language pattern processing for data machine translation; "In other words, Google's tasks are similar to what the brain functions: learning, classification, vision, and language."

This series of tasks provides an analysis of a great deal of information-from voting habits in the U.S., fluctuations in airline fares, to a wealth of health data. "The biggest challenge in the petajoules byte era is not to store all the data," the Wired magazine said in a recent commentary, "but how to understand the data." "Understanding the data, this is the current development direction of internet intelligence."

Internet--A "cloud" resource

What is the connection between the Internet and cloud computing? The internet is a "cloud"-a large number of computers connected together to communicate. The internet is not primarily about computing, but nothing can stop us from making the entire Internet a resource pool with huge computational power. If you combine the Internet with cloud computing, you will have a complete machine that accomplishes three important tasks in the brain: information storage, processing, and communication.

When the information processing cloud model and the communication cloud model are combined to run in parallel with a circuitous and gradual way of thinking similar to the human brain, the same wisdom may appear on the Internet. In the 2003, a few years before joining Google, Google's president, Eric Schmidt, said: "When the network becomes as fast as the processor, the information in the computer is released and spread across the network." "At this stage, all the information on the Internet must be compressed, just like our paper-based brain model; At this stage, a pattern must be established, at which time a lot of ideas are fleeting, and then intelligence is generated."

In the latest issue of The New York Times, Bill Gates believes that the development of technology--to smart development--is as unstoppable as the salmon. "They (Google) do a distributed computing system. He said, "but our dryad will do better." "He paused, and then had to admit that there would be more powerful, more suitable systems to appear," and they would do better. ”

Bill Gates is right that development is unstoppable. Have we changed the Internet, or has the internet changed us? "The first article, the author: [US] Jeffrey Steeber, the article title and the text title for editor Plus, Ashing finishing)