2. Extensive network access: It means that the resource we want to use can use any network to connect to this resource
3. Resource pooling: resource pooling needs virtualization to achieve
4. Fast and elastic scaling: that is to say we can apply for services on demand
5. Measurable services
2. The definition of cloud computing:
The National Institute of Standards and Technology (NIST) defines: Cloud computing is a model that can obtain the required resources (for example, network, Servers, storage, applications, and services), resources can be quickly supplied and released, reducing the workload of managing resources and interaction with service providers to a minimum.
3. Development of Computing Technology
1. Serial calculation: Traditionally, general software is serial calculation, that is, a "problem" is divided into a series of discrete "Instructions", and each "Instructions" will be executed one by one on a single CPU, and the CPU is in the same Only one "Instructions" can be processed within a time
2. Parallel computing: Decompose a problem into multiple parts that can be processed at the same time, and then divide each part into a series of discrete instrutions. Each part hands its own instrutions to its own CPU for processing, and each Cpu simultaneously Accept the instrutions of different parts, and then add a unified control mechanism to control the entire process
3. Distributed computing: Distributed computing belongs to the field of computer science that studies distributed systems. A distributed system is to disperse all its components on computers belonging to different networks, and these computers communicate and cooperate with each other through a unified message mechanism. Components distributed on different network computers cooperate with each other to accomplish a common goal.
The difference between distributed computing and parallel computing: If the processing unit (processing unit is what we call CPU) shares memory, it is called parallel computing, and vice versa is distributed computing. Some people think that distributed computing is a special case of parallel computing.
In fact, the distributed task packages are independent of each other. The result of the previous task package is not returned or the processing result is wrong, which has little effect on the processing of the next task package. Therefore, the distributed real-time requirements are not high, and calculation errors are allowed (because each calculation task is calculated by several participants, the results must be compared after being uploaded to the server, and then the results with large differences are verified). The task packages processed by parallel programs are closely related to each other, and each task block of parallel computing is necessary. There is no wasteful segmentation, that is, each task package must be processed, and the calculation results affect each other. , This requires that each calculation result is absolutely correct, and try to be synchronized in time, and many distributed task blocks can be processed without processing, such as a large number of useless data blocks, so the speed of distributed calculation is very fast , But the real "efficiency" is low and low,
4. Grid computing: Grid computing uses a wide range of scattered computing resources to complete a common task, and it is also a type of distributed computing. According to IBM's definition of "grid", it aggregates scattered available computing resources on the local network or the Internet, so that end users or applications feel that they are using a powerful virtual computer. The vision of grid computing is to create a virtual and dynamic collection of resources so that individuals and organizations can use these resources in a safe and coordinated manner. Grid computing is usually implemented in clusters.
4. Classification of cloud computing by operating model
1. Public cloud: built by a particularly powerful vendor, because public cloud needs to deploy data center nodes around the world, large service providers provide their own resources to users, for example: AWS is a public cloud business provided by Amazon
2. Private cloud: that is, a cloud built by a company itself. Alibaba and Tencent do not provide private clouds. Huawei provides private clouds. Of course, there are some other companies that provide private clouds.
3. Hybrid cloud: Hybrid cloud is a relatively flexible cloud computing model. It may include two or more of the public cloud, private cloud, or the industry cloud to be discussed later. The user's business can be based on demand Switch between these types of clouds.
4. Industry cloud: transportation industry cloud, financial industry cloud, pharmaceutical industry cloud, etc.
V. Classification of cloud computing by service model
In cloud computing, generally all the applications we deploy follow a unified hierarchical structure. The application is finally presented to the user. The user saves or creates his own data through the application interface. In order to ensure the normal operation of the application, it is necessary Rely on the lowest level of hardware resources, the operating system running on the hardware resources, and the operating environment of middleware and application programs running on the operating system. We call all parts including applications the software layer, and call the lowest-level hardware resources, including network resources, storage resources, and computing resources, as well as the virtualization layer, the infrastructure layer, which runs on top of the operating system and applications. All the middle parts below are called the platform layer.
1. Traditional mode: Generally speaking, it is a chimney type, and all components need to be independent
2.IaaS (architecture and service):
If the infrastructure layer is provided by the cloud service provider, and the others are operated by the user, this model is called laaS
3. PaaS (platform and service):
If the infrastructure layer and platform layer are provided by cloud service providers, and the others are self-operated by users, this model is called PaaS;
4. SaaS (software and services): If all are provided by cloud service providers, this model is called SaaS.
