4 issues to be faced with the rosy outlook for cloud computing

If we look back at the 2008-year IT sector, the most memorable of all, in addition to the impact of the global financial crisis on the IT industry, I am afraid the rapid rise of the non-cloud computing. Starting in the second half of 2008, a set of forums and exhibitions on cloud computing have been held, and the industry's focus on cloud computing is far greater than any other technology. But, as more and more people are keen to discuss how to migrate to cloud computing, there is concern: If users are dissatisfied with cloud computing services, how to exit from cloud computing, and how to prevent the vendor from being locked out. The consultancy survey found that fear of being locked in by suppliers has become a very important consideration for many IT leaders in deploying cloud computing applications.

As a new IT model, cloud computing enables businesses to use various http://www.aliyun.com/zixun/aggregation/13543.html ">it services over the Internet without having to purchase, deploy, and configure various IT systems themselves, Thus the enterprise from the IT system maintenance work freed up, and put more energy to their core business. Thanks to this feature of cloud computing, more and more enterprises, especially those with weaker it technology, are beginning to consider migrating their IT systems to cloud computing. And experts remind companies not to be locked by the manufacturer, the migration must be carefully evaluated, and to save themselves a good retreat, in case of a dilemma.

Currently, there are vendor-related reasons for this lock-in (intentionally, for example), as well as technical reasons. For example, the lack of standards for cloud computing is one of the most important reasons, because there is no standard for data and applications to migrate freely between different cloud services providers.

James Staten that this would leave a hidden danger for future migrations. If one day the user is dissatisfied with the services of the cloud service provider or the supplier is no longer providing services for a variety of reasons, users must migrate to other cloud computing platforms, data and applications may have to be modified, which is often a complex and time-consuming task.

However, Staten that "locking" should be a dialectical look, he said: "As long as you deploy your application to the cloud, in a way, you put your own application and cloud suppliers together." For example, you might use a vendor-specific virtual machine or API, or other unique configuration and combination methods. ”

In fact, the more proprietary features that users use cloud computing platforms, the more likely their application functions will be, and the higher the user will be locked in by the vendor. Therefore, it is entirely possible that even with the same vendor and the same cloud computing services, some users will be locked by the vendor and some will not.

The technical challenges of cloud computing

工欲善其事, prerequisite. Although the future of cloud computing is beautiful, there are still many technical obstacles to be solved, including highly reliable system technology, scalable parallel computing technology, massive data mining and data security technology.

1. Highly reliable system technology

It is a large-scale cluster computing system that supports cloud computing, and reliability and stability become one of the biggest challenges when the scale of the systems increases. Through effective system configuration, monitoring, management, scheduling, virtualization and other technologies, to achieve a strong, dynamic, autonomous computing storage resource pool, to provide cloud computing needs of the large capacity computing power.

System-Level fault-tolerant technology is a difficult problem in system technology. When a large number of servers perform the same calculation, a single node failure should not affect the normal operation of the application. For similar data retrieval, it is easy to compute the application of communication between nodes. However, there is still no effective system-level fault-tolerant scheme for tight coupling applications with a large number of communications. At present, mainly rely on the application level of checkpoints and restart technology, on the one hand, increased the difficulty of development and workload, on the other hand, the performance of the operation also has a certain impact.

2. Scalable Parallel Computing

Parallel computing is the core technology of cloud computing and one of the most challenging technologies. The presence of multi-core processors has increased the level of parallelism, making the development of parallel programs more difficult than ever before. But the current industry does not have an effective parallel computing solution, whether it is programming model, development language or development tools, distance from the expectations of developers have a big gap. The automatic parallelization solution has been proved to be a dead end in the past 30 years, but the traditional manual parallel program development method is difficult to be mastered by ordinary programmers. Intel, Microsoft, SUN, Cray and other industry giants are investing a lot of manpower and resources to carry out the relevant research, but the real mature products are very difficult to appear in the short term.

Scalability is one of the main considerations of parallel computing in the cloud computing era, and the application can be expanded effectively with the increase of user's request and system scale. At present, most concurrent applications are difficult to achieve effective acceleration performance on more than 1000 processors (cores), and many future parallel applications must be effectively extended to thousands of processors. This is a huge challenge for developers.

3. Mining technology of massive data

Cloud computing is faced with terabytes and even petabytes of massive data, how to obtain effective information from the data, which will be the key to determine the success or failure of cloud computing applications. In addition to using parallel computing technology to speed up data processing, new ideas, methods and algorithms are needed to accomplish more accurate, fast and powerful data mining.

In addition to the massive data mining, massive data storage and management will also be a huge challenge. In the era of cloud computing, databases will face a serious crisis, either by effectively expanding the cluster database to thousands of nodes or by replacing it with new technologies like the Google file system. "Database is dead", this assertion will become possible.

4. Data security Technology

It is not easy to change the data that was originally kept locally and in the hands of an external cloud computing service Center. With the development of network technology, bandwidth will not become the main obstacle, security is still the most important concern. However, as has been accustomed to the money in the bank, the future of the data banks will appear, but only a matter of time. Technology is not the main obstacle, systems, regulations, integrity, habits, concepts, these non technical factors will determine the popularity of cloud computing.