Light Zheiyun: Graphic cloud computing (fun and easy)

The concept of

Cloud computing has been rapidly sweeping the IT industry since its inception, and the industry's giants have been massively killed in the cloud of "the word will be called cloud computing". Oracle CEO Ellison said: "If there is any industry than women's fashion industry more trendy and fashion, it is not the computer industry mo." "More industry bosses wear clouds like the neon feather clothing on the cover of fashion magazine Bo Jun a smile, but no one has ever seen cloud computing in the neon light after the appearance of the solution." This article will lead you into the cloud, a few illustrations clear cloud computing.



A: "What is cloud computing?" Who gives a simple explanation of the popular point ah ...

B: "Probably like a network hard disk can upload computer backup and file can also edit the online I what, nothing special ah is not already realized, only feel the big head now is want to put them under the plan." "

A:" Khan. That's called Network hard disk ... "


What is cloud computing?


Figure 1
  at the very beginning, people use an abacus.


Figure 2
Later, people use computers.


Figure 3
Later, people have a network.


Figure 4
Later, China's large population explosion, male and female ratio of men more than 37 million, the more than 30 million people have nothing to do, all go online. So the server is too overwhelmed.


Figure 5
   so people invented the technology to use better servers.


Figure 6
Later, more people, and more servers.


Figure 7
But in fact this is not a good effect, the excessive structure of the Web site design and construction more difficult, and the more complex the more unstable the system. There may be a problem, such a complete system will be completely hung up. If you take into account the collapse of the system, there is a need to introduce a more complex scheme to ensure that different servers can do different support. This is a no solution of the cycle, a large number of computing resources are wasted in the infinite entanglement, quickly to the bottleneck.


Figure 8
People think, I don't need to be so messy and complex system, my last great server is not good? But it's too expensive ... And the best is not yet made.


Figure 9
A good idea suddenly came to mind: to assemble all the computing resources as a whole (a cloud) to complete the operation request through concurrent use of resources. Each operation request can be divided into small fragments according to certain rules, distributed to different machines at the same time, each machine in fact as long as a small calculation can be, this is even 286 of machines are easy to complete. Finally, the computing results of these machines are integratedand output to the user.

to the user, he is not in fact facing many machines, but a seemingly real computing power of a single server, more than 10 system z10 mainframe rub together, or to create a new era of petaflop "Dawn" and "Sequoia" number and cattle. In fact, this server does not exist, but it has the ability of thousands of servers.



To see the example below.


The process is not so simple. Even the process of collecting data can be a terrible processing time. This will further divide the task of cloud computing, which server CPU what to do, processing which task segment. This can actually be arranged by the algorithm to automatically allocate.

In summary, squeezing the potential of each step, let a task by the server cluster together, naturally can be achieved quickly. Don't forget, cloud computing is not a two or three-quality Low-cost server can be achieved. Behind each cloud is a heterogeneous platform server, especially a "private cloud" built into the corporate firewall. Because the enterprise's computing needs are often complex, choosing a different platform to deal with different computing needs is the most cost-effective, which is the same as what the farmer uncle pulls or chooses what car. Fresh Chinese cabbage preferred motorcycle, fresh and quick to use blades; advanced Chinese cabbage preferred reliable transportation power server; A large number of high-grade cabbage choose a big truck, just as the I/O throughput data is suitable for large host system z, is it faster than a two trip to the cart? Large farms are not confined to one plant just as large enterprises do not have only one computing need. So there is responsible for central management, monitoring software Tivoli, difficult to use the human brain statistics? Finally, the farmer's uncle (the very bull's it guests) assembled the teams to form a hard part behind a cloud. Few it bosses can set up a full set of models, as far as I know it seems only IBM can.

The next solution to the softer problem is: when the base of existing computing resources is unchanged, cloud computing makes the user's task request division, a request comes in, we turn it into a lot of small task segments, and finally summarizes the user to a complete result. To the user, he simply does not feel the inside which CPU does what processing, which part is which part stitching up, he feels oneself to face a 500 million memory 300 million GHz Big Mac computer.

The user was baffled by this calculation, and he called it cloud computing.