Qiping: Cloud computing is transforming production

Now many industries or services are in the banner of cloud computing, but not all of it is cloud computing, quweicunzhen see cloud Computing, is to capture the real core of cloud computing,

There are two important aspects of cloud computing, neglected, one is centralized computing, the other is distributed computing, cloud computing is a specific way to integrate the historical products. Ignoring any point will have serious one-sidedness.

Now a lot of "pseudo cloud computing" is not necessarily said to be completely different things, but to the cloud of a side, or a local source of history, exaggerated to the mainstream of cloud computing, and produced misleading.

To this end, here is intended to only the core content of cloud computing, to conduct a discrimination.

1 centralized computing is divided into several cases, and cloud computing has a connection, and there are differences:

The first, and most typical, centralized computing is mainframe, such as IBM's s/360. Quite a few people think of cloud computing as centralized computing, which is not comprehensive. Because cloud computing is obviously not equal to mainframe. The application of cloud computing is not concentrated in the mainframe, but is distributed on the external compute nodes of the centralized computing and storage.

The second, is the master-slave structure, the server is the master-slave structure of the focus of one side, the internet after the web model, is a master-slave machine structure. Both the master-slave structure and the web are not equal to cloud computing, because their control is on the host (server), while the machine is decentralized, but without control, not user-centric.

The third is grid computing, and grid computing is still centrally computed. Although it has the characteristics of using distributed computing resources on the surface, it is not distributed from the application point of view because of the huge and single computational problem to be solved. By understanding cloud computing as grid computing, it is clear that the node (application) task is presented.

2 distributed computing, but also with the cloud computing has links, but also different:

The first is that the internet before the advent of the Web (the Internet before WWW), such as the Apache network, is typically distributed computing and does not require a server as a computing center. Distributed computing alone does not equate to cloud computing, because cloud computing also has a centrally stored and calculated portion.

The second, Peer-to-peer, is that a computer is directly connected to another computer without a server. Its important feature is reciprocity, there is no master-slave. In a manageable peer-to-peer mode, although you can go through the server, but different from the master and subordinate machine structure, it can also not through the server, and the master-slave structure is not through the server is not possible. Cloud computing differs from Peer-to-peer in that it must focus on computing and storage as complementary aspects of distributed computing.

Third, collaborative computing, can be considered as peer-to-peer form of the promotion, from point-to-point, to group collaboration, instant messaging, SNS, can be based on collaborative computing and parallel computing. Collaborative computing is very close to cloud computing, but there are still differences, such as SOA (service-oriented architecture) that can reuse service modules, but this may only happen between applications, and it must move these reusable resources to an application-independent cloud server for cloud computing.

Many of the existing definitions of cloud computing tend to favour one side of the above, and therefore seem specious. It should be reminded that the industry talk about cloud computing often has a specific context, such as cloud computing to strengthen centralized computing and sharing. The combination of specific points may not be wrong, as it may refer to the construction of the cloud infrastructure or platform itself, or to the cloud business of a large enterprise. But listen to people often hear wrong, think this is the whole cloud computing, not realize the distributed computing and application side of the existence. This leads to misleading and misunderstanding.

This is not a scary situation, and the elimination of misunderstandings will make it clear. But there is a different situation. Some large enterprises in the cloud computing, with the traditional grid computing, the old idea of the master-slave computer, the cloud into the exclusion of distributed computing in the central calculation of the track, it belongs to the expert tradition and conservative. There is a real problem with the understanding of integration at present.

3) A concentrated and decentralized connection

Many of the existing definitions of cloud computing overlook the connection between two basic points. In fact, the connection between the two is the essence of cloud computing. This kind of connotation, also is from the history step by step develops. There are several concepts that deserve special attention:

First, share (or share). Now all cloud computing will mention sharing. This mainly refers to the sharing of reusable resources in the cloud, including the sharing of point-to-point. Sharing is a concept that is incompatible with industrialization. The Industrial enterprise economy is established under the premise of excluding sharing; What can be shared belongs to the public goods provided by the Government in tax revenue. But in cloud computing, sharing can be a market behavior. The reason for this is that there are two situations that have changed compared to the industrial economy, first, the means of production from the material and capital, into the data, never reusable, into reusable; Secondly, there is a cross subsidy between the basic business and the value-added service, and the share can be rewarded (for example, free mode). This results in concentrated partial sharing, while decentralized partial returns (in the Apple App Store model is 37 divided, by 70% for distributed developers, and 30% for shared tools and platforms). Among them, the reusability of data resources is the focus of cloud computing, the basis of using sharing mode, and the connection between centralized business and decentralized business is the basis for sharing the model that can be established. Many people talk about cloud computing when sharing, only to focus on the sharing of resources, but not linked to the return mechanism on the node, the cloud is confused with learning Lei Feng good thing, the equivalent of the open source software and free software mix. This makes the infrastructure of cloud computing a people's commune, becoming unsustainable or something else.

Second, virtualization. Virtualization is an important feature of cloud computing. The mode of industrial production is concentrated and the agricultural production mode is dispersed. History never solves the contradiction between the two. Virtualization in the coordination of the relationship between the two, out of a new road. As early as the mainframe era, the idea of virtualization was put forward. But there was no end to the application. In cloud computing, virtualization is a key link between the concentration of the cloud and the dispersion of the end. Actually, technically, the framework of capital specificity has been broken. In the real economy, the elements are rigid, it is difficult to free flow, fixed assets investment, often become sunk costs, resulting in huge idle and waste. In cloud computing, virtualization actually turns the rigidity of the elements into flexibility. This is not a direct change to the marginal input of application and value-added, but a huge social saving for the shared resources that focus on the cloud. The equivalent of a fixed cost (such as initial fixed input) for the same production saves a lot. Society can turn fixed costs into variable costs, focus on the value-added services necessary to create more value for users. The more far-reaching implication is that when this approach is combined with the scope economy, the "small farmer", characterized by personalization and customization, is resurrected (certainly at a higher historical stage, on the basis of more advanced productivity), and the contradiction between centralization and decentralization is discarded.

Third, utility calculation. Utility calculation is actually the concept of leasing. Because the resources can be reused and the factors of production can be virtualized, a reasonable change in the property right is that the resource allocation can be "changed to have control (use)", separating the ownership from the right of use, not having ownership of the factors of production, and using the way of leasing to control the use of rights In cloud computing, in this way, it is the natural connection between the ownership of the cloud side and the right to use the end side. This creates a curious mix of public ownership and private ownership: In the cloud, infrastructure (I), platforms (P) and software (S) are "public" in the industrial age, and in the end, value-added services and marginal inputs are "private", while "public" and "private" are divided in a certain proportion ( such as 37 split).

The Open API (Application interface). When it comes to defining cloud computing, the API is rarely taken into account. In fact, between concentration and dispersion, leaving the API connected, everything becomes unworkable. So it can be said that the API is the key to cloud computing, or the cloud and end-link mechanism, is the key to cloud computing. In the industrial economy, the API does not exist. Even in the Internet economy before the advent of cloud computing, APIs are often closed or have no real effect. For example, in SOA, shared modules are not independent of the API form, provided by independent third parties (like the cloud). Many people who talk about cloud computing do not share their resources in the cloud, and how to compute connectivity with terminals as an important aspect of cloud computing.

I personally think that cloud computing in a deeper sense, is to improve efficiency (that is, pay increase) as the goal orientation. Many people simply talk about the problems that cloud computing can solve from efficiency, including cost savings, and are not fully in place. It can be falsified to improve efficiency, or it can be diminishing returns, but in that case the cloud (I, P, S) is no different from the cloud (physical infrastructure, plant, and equipment). The real difference can not be summed up without talking about efficiency. The real difference is in efficiency, the rate of change, that is, the more ... More ... "): Using cloud computing, the more complex the system, the lower the marginal cost; without cloud computing, the more complex the system, the higher the marginal cost. In this sense, the adoption of cloud computing is essentially a shift in production.

(Responsible editor: admin)