Cloud computing needs to face six challenges

After a long period of development, cloud computing has been recognized by many customers. Here I publish my personal opinions and discuss with each other. 1. Security: Cloud computing security has two main aspects: First, the security issues brought by the cloud computing environment, the second is how to achieve cloud computing security protection.

The former includes technical aspects of the security issues and policies above the regulatory safety issues. The latter is mainly reflected in the traditional sense of the security software systems are built on their own loom based on these software services in the framework of the cloud environment, the software itself is far from the security requirements of users, you need All software vendors change the original safe mode of operation, consider the safety of their products in the cloud environment and protection mode.

2. Availability: Availability refers to the software system in a given period of time to work the proportion of the total time. In order to improve the really good service available, cloud computing providers are studying common fault analysis and prediction models. Based on the research of these models, cloud computing service providers hope to predict the possible availability problems and avoid or reduce these failures by preparing ahead of time duplicates, solving problems in advance, notifying users and so on.

3. Scalability: Scalability software systems can respond to load changes with increased or decreased resources while maintaining consistent performance.

In a cloud computing environment, both vertical scaling and horizontal scaling of applications are supported through the cloud computing infrastructure platform. For example, in a server-based virtualized cloud infrastructure, vertical scaling can be achieved by adjusting resources of a virtual machine. For horizontal scaling, the virtual machine node corresponding to the application can be increased or decreased in the same manner.

4. Confidentiality of information: In a cloud computing environment, although the complete data is fragmented and stored on different servers, it increases the difficulty for the illegal users to obtain complete data, but it still may exist. For example, illegal users can violently crack all the storage servers to collect information, he will even crack the cloud storage system data distribution logic. In addition, we must also consider the legal policies of various countries.

5. High Performance: We focus on analyzing the performance of the most popular technologies in current cloud computing environments, including server virtualization, mass data processing, and distributed storage.

In today's popular paravirtualized systems, such as Xen and VMware VMware, the hypervisor brings only a small amount of additional CPU overhead, while memory and I / O are more expensive. For the current virtualization technology, the original CPU-intensive applications can better migrate to the virtualization platform, and the original memory or I / O-intensive applications, such as databases, will encounter a larger trouble.

As a de facto standard framework for large-scale data processing in cloud computing, MapReduce also has performance issues. The first is the performance issues due to applicability, because Google designed MapReduce for search engines, not entirely from the common starting point to consider. Second, MapReduce primitive design can also cause performance problems. In addition, since MapReduce runs on a distributed system, nodes in the system are connected through the network, which requires a lot of network message communication during operation, resulting in additional communication overhead.

6. Standardization: If users want to maintain data synchronization among multiple clouds, synchronize application versions, or interoperate with cloud applications, it is best to abstract multiple cloud data centers into one method in a single way. Once to reduce the complexity of use. This job function is homogeneously standardized.