"Live Report" centralization and distributed architecture

Good morning everybody, just now Mao always from the strategic point of view of the future of cloud computing and development direction, I intend to from the angle of engineering or implementation. It should be said that from the perspective of cloud computing, we have a lot in common. From the engineering and implementation point of view, how do we achieve a cloud computing. Is it centralized? Distribution？ or balance? So, how to achieve a sustainable development of a cloud computing architecture?

First, let me conclude by looking at what the cloud is going to look like from Intel. Dr Mao has talked a lot about future visions and the benefits that cloud computing brings to everyone's lives. Intel is a technology provider and a manufacturer of cloud computing ecosystems. Last year, we released our vision to the world. 2015, the world colorful clouds. They are federated and can be shared between the data. Again, we'll have a big data center. These data centers may require remote access, so these data centers must be managed automatically. This is important because when you have thousands of servers hundreds of of miles away or even thousands of kilometres away, you are not impossible to manage by people, but need a set of technical tools to manage, this is the future development of cloud computing, a feature of the data center. Another feature is that cloud services must be perceived by the client. What is client-side perception? Client-side perception that the services provided must be personalized, local conditions, INN, due to equipment, I would like to perceive that you are in a kind of network environment, I provide you with services. Who you are, I give you the appropriate equipment. Because the future use of cloud services or access to cloud services a wide range of devices, such as: television, netbooks, tablets, traditional PCs, smartphones, are the terminals of cloud services. Therefore, the provider of cloud services must be personalized-based on equipment, the network, the user's situation to provide the corresponding services this is our vision of the cloud, summed up, that is, cloud and cloud to unite, cloud management to automation, cloud services to personalization.

To achieve these visions, Intel has several specific steps:

First of all, we have a good relationship with users, the industry, cloud computing Open Data Center Alliance ODCA is an Intel-sponsored user Alliance, currently, it has more than 200 end user companies to join, they represent 70 billion of dollars in revenue. Intel is a sponsor and technical advisor to the Alliance, not a member, because we are the manufacturer. Last week, the alliance just released its first version of the road map of its members ' common usage model, a total of eight, "What are the common characteristics of the future development of cloud computing data centers", they send these development models to industry and manufacturers, to standard-setting people like DMTF, OASIS, they can analyze these patterns, See what criteria are used to support these patterns. Second, with these patterns, we will be in the future in the data center, we use these models with Huawei, Wave, IBM work together to solve these problems, and then produce a reference framework. So we have a cloud computing program to help vendors who want to build cloud computing. Such as: You want to do green data center, that green data center have what solution, we help you build.

The client is the equivalent of standing on the shoulders of giants to make new innovations, so, from the cloud's strategic point of view, we focus on three pieces: a vision to support the cloud from products and strategies; second, work with end users to understand and agree on what our future data centers are in the context of cloud computing, what are their common needs, What are the common patterns of use? We work with vendors to support the needs of these users. Specifically speaking, the demand is changing, this is the consensus of the industry, and this thing will certainly happen in the next few years. One feature, the future content of cloud computing is important. Video, data, etc. to account for more than 90%, we are very interested in how the media cloud to provide users with better and more effective services, how to more users, more data, more devices in the environment to provide extended services. So, this is a very challenging proposition. When a small technology benefits the small crowd, easier. But when it comes to hundreds of millions of people and a wide range of land, such services must be scalable. This is a problem that it people have to consider. are now a large data center, the greater the bandwidth of the network the better. Perhaps, the future of the data center, as long as there is a browser, there is nothing. This can be a model of cloud computing, but whether it is the best model, whether it is a sustainable model. So, this is a question to consider.

Let's take a look at one of the major challenges to implementing these visions of cloud computing, which exists at all levels. Future equipment is diverse. At present, the management of these devices is very poor, especially just speaking, to the user to perceive your service, you must first perceive the user's device, the user's environment to perceive. Now, this perceptual means of managing these devices is relatively poor. Second, the user's expectation is that I don't have to consider the cloud's computational and storage limitations, which are extensible. Users want this service to be available anytime. This is not the same as the way we used to sell hardware or software. Data center This side, China to serve 1 billion people, the world has 2 billion netizens. In these large services, scale, security is a problem. Multiple leasing means that the data center is so large, different users may share a server, sharing a storage space, sharing a data center, then how to ensure the security of data information? Because, in the era of cloud computing, as an individual, it is impossible to enjoy the exclusive server, or a unique environment, the cloud era of the data environment is shared, then the privacy security is more important, then how do we manage the model of multiple leasing. Third, regarding the Green data center, the traditional IDC, the electricity tariff and so on accounted for is quite high. How to reduce operating costs? These are the problems that we have to consider in the framework of cloud computing.

This is one of our cloud computing architecture, which is that I am telecom, my data center in Nanjing, but my users may be in Nantong, so I need to connect to the user through a network system, I in the provision of services, is a centralized service mode, the network is simply the transmission of data, intelligence reflected in the data center, Then your service, user perception is done in the data center, many users are in the home, residential, building access to these services, this is a pattern, but it for the data center and network requirements are relatively high. We raised the idea of whether the cloud can be distributed, some of the functions of the cloud distributed to the edge of the network, or distributed to the community, we call "Community cloud server." Then, the function of the cloud to the edge of the network, and then you to the user's network advertising push, etc. can be pushed to the edge of the network, which will reduce the central data center and network pressure on the edge to provide instant service. The quality of service will be better. The Master of all content will still be in the centralized data center. and video services, personalized services (push ads) and so on, I can directly push advertising services, so, from this perspective, is a balanced cloud computing model. This is complementary to the centralized model, which may make it easier to solve extensibility problems when it comes to community outreach. The equivalent of such a small data center, the center of the data center and network pressure greatly reduced. This is a concept that can be solved technically. There are also business models, such as the integration of cloud service providers, and so on, which remains to be solved. Today, the concept is thrown out, is to communicate with you, a good idea.

Cloud, in fact, many are end-to-end problems, just I am talking about the cloud backend problem, need data center management, design framework. The need to consider distribution or concentration, and then on the client also need to consider, in different devices also need common management standards, your device's cognitive, this is difficult to achieve. This can also seriously affect the user experience. So this is a system of engineering, not a certain point.

That's all I'm saying today, thank you.

(Responsible editor: admin)