Subscribe cloud computing RSS CSDN home> cloud computing

Cloud computing platform to build and practice Forum (two): CDN, Docker into the focus, Intel, but also to take the cloud lift to redefine the trend

Published in16:19 2015-06-08| Time reading| sourceCSDN| ZeroArticle comments| authorWei Wei

Abstract:On the afternoon of December 5, cloud computing platform construction and practice forums, from hyper, Intel, China Telecom, Hangzhou and shoot cloud, Shanghai Sheng Amoy data a total of six experts, gathered and share dry, everyone virtualization, docker, openstack, CDN, cloud acceleration etc. has carried on the thorough discussion.

[CSDN] 3-5, June 2015, as the most influential event in the field of cloud computing:The seventh China Cloud Computing ConferenceGrand opening at the National Convention center. Conference to promote the cloud computing innovation and development, cultivate the new format in the information industry "as the theme, inherited the former sixth successful experience, more than 100 domestic and foreign cloud computing core and experts in the field of wonderful speech, focusing on ecological system present situation and the trend of development, exchange of experience, further pushing technology innovation and application.

On the afternoon of December 5, cloud computing platform to build and practice forums, founder &CTO of for the hyper Wang Xu, Intel data center cloud platform division whole cabinet architecture market, technical manager Ruan Xingping, Intel China Shanghai branch company data software engineer Luo Zhongyue, China Telecom shares Co., Ltd. cloud computing company technical director wide Xiaoming, Hangzhou and take cloud CTO Huang Huipan, China, member of the Committee of experts on big data, Shanghai Sheng Tao Big Data Technology Co., Ltd. General Manager Ding from their perspective, cloud platform construction practice and share their experiences.

This field in the China Electronics Association Committee member of the cloud computing, the Chinese Academy of Sciences, cloud computing industry technology innovation and Incubation Center Director, Ji Tongkai, President of the Guangdong Institute of electronics industry under the auspices of the official start.


Deputy director of the Committee of China's big data experts, China electronic society, vice chairman of the Committee of experts on cloud computing Huang Xiaoqing as chairman of the forum delivered a speech. He said today several speakers is cloud computing industry, which we are all concerned about the latest technology introduction, now cloud computing infrastructure faces two important issues, a is network performance problems, a is infrastructure of storage cost. This will involve the problem of service costs. Both to ensure security, but also to ensure the quality of service, but also to ensure that the cost is low. Today's forum is a good opportunity, and finally he hoped that we can be based on their specific problems encountered in the actual work and the guests we talk today, together for effective interaction.


HYPER founder &CTO Wang Xu: Hyper - based on the virtualization of the Docker engine

HYPER founder &CTO Wang Xu said in his speech, is currently in the Docker boom, what is Docker? He thinks Docker=Container (LxC) +Layered image (aufs), both LxC and aufs is not something new, and combination has produced great effects. The reason is that the docker's core is used as the center, to greatly simplify the overall release process, reduce the difficulty of operation and maintenance. Docker for the application of the package, so that the operating system back to basics, application and operating system completely decoupled, is a revolution in the field of operation and maintenance. At the same time, he said virtual machine is not do the application as the center and hyper is center application virtualization. After that, he also introduced the nature of hyper, and has a container, why also virtual machine and hyper and containers were compared.

With Docker than up, the virtual machine is not to do its problem to be applied as the center. And Image Hyper, Hyper is not the same point, HyperVisor to HyperVisor as the engine, in the Run inside Docker Image Hyper=Hypervisor+Docker. So I will continue to uphold the previous HyperVisor brought from isolation to a variety of advantages and compatibility above, but also can be done as Docker such as the use of a center for the use of.


Hyper running architecture is such, it will use a lightweight Initrd under the hyper, and then directly to the docker image runs in the top of the Initrd, use method and docker is exactly the same, with the docker down the image push, and hyper HyperVisor in operation. The Certainly a lot of people are thinking about a problem, HyperVisor is not will become very heavy, HyperVisor virtual machine has changed very heavy, because it is too much emphasis on the traditional hardware simulation and run a full operating system. As a reasonable reduction, in fact, can achieve a performance close to Container.

Big data Companies Software Engineer Luo Zhongyue Intel data center cloud platform division whole cabinet architecture market manager Ruan Xingping & Intel China Shanghai Branch: Intel whole cabinet architecture for software defined Infrastructure Optimization of cloud infrastructure solutions

Intel data center cloud platform division whole cabinet architecture market, technical manager Ruan Xingping published entitled "Intel machine cabinet architecture: for software defined Infrastructure Optimization of cloud infrastructure solutions", he said with the development of cloud technology to construct data center requirements become more and more standard, deliver more elastic, the deployment of more flexible. Intel's entire cabinet is a logical architecture for computing resources, network resources and storage resources to achieve decoupling and the need for restructuring. This scheme has been applied to redefine today's data center cloud architecture so as to realize the efficient deployment of cloud services. Finally, he also shared the combine software defined infrastructure case with openstack ironic and Intel whole cabinet architecture.

The goal of this logical scheme is expected to be able to reduce the cost of assets by 25%. At the same time to improve the productivity of each unit of IT capital. The most important part of the cloud business on-line deployment time, can be very fast and effective implementation.

Next through an animation, Ruan Xingping introduced the whole cabinet architecture. First of all, the first of them is the pool of hardware resources after the decoupling. In the resource pool system, can according to different applications, the reconstruction scheme, in the whole cabinet, storage, network, computing resources to obtain related resources, construction of a physical logic machine. In the above can make an abstraction of it that this is a cloud physics, from the inside out resources, and then through the module of multi cabinet management module Pod-wide management and the need by efficient interconnection network is the whole cabinet structure inside each group of interoperability. At the same time put forward the proposal of a multi cabinet, in the entire multi cabinet inside, these stores are shared to use.


Intel China Shanghai branch of the big data software engineer Luo Zhongyue Ruan Xingping speech, from a technical point of view of Intel's entire body is how to combine the OpenStack. Mainly includes two steps, (1) RSA in the OpenStack rendering; (2) the use of RSA to expand the function of OpenStack.

Openstack compute model, there are mainly three kinds, a kind is compared with traditional provide virtual computing services, the middle is to provide physical machine computing service model, and a kind of is that you can use openstack to do a cloud based, on the basis of the cloud can build some dedicated cloud. The basic idea is RSA mechanism is a set of management software, this set of management software provides the API, the API and converted into corresponding openstack components inside the driver, provides an interface for a complete set of technical architecture, the openstack is based on RSA software management provides interface to a set of standard of openstack, so let the RSA openstack the.


China Telecom Co cloud computing Branch Technical Director of Guangzhou Xiao Ming: China Telecom cloud computing research and Practice

China Telecom Co., Ltd. cloud computing points the company's technical director Guang Xiaoming published entitled "China Telecom cloud computing areas to explore and practice," the speech, he said China Telecom has set up a North and south two cloud data center, covering the whole country to build eight large area, 2 cloud base, 2 large isolation zone, a total of 17 resource pool node (not including the isolation zone and two base) and share the Tianyi North Yunnan double live storage pool case and hybrid cloud application case. At the same time, he also focuses on the hybrid cloud Tianyi five scenarios include disaster recovery, has elasticity, test, research and development, mobile app and virtual desktop. In the end, he also introduced the core technology index of sky wing hybrid cloud.

Guizhou China Telecom park officially put into operation, China Telecom completed four years of localization of the South and the north two cloud base of production and construction. With Guizhou on the line, the north of the Inner Mongolia base and the south of Guiyang, your new district base officially opened commercial production. Of course, for China Telecom, it is some of the IDC how to do cloud transformation, this is actually a very prominent issue facing China telecom.

Throughout the public cloud from the income scale, it's accounted for a small part of the. In addition to this is a hybrid cloud. This part of this enterprise was originally built, or it has been cloudy, then this part of the work, it would have to be managed out of resources how to perform a better interaction. This is in fact the original intention of the entire network of China Telecom. Then from the point of view of Telecom Tianyi cloud, in fact, scheduling faces two challenges is the need also to scheduling and arrangement of the so-called public cloud resource pool and hybrid cloud resource pool, that part of the private cloud is to accomplish alone. In this process, for telecommunications. It is a big challenge is from the original enterprise, it built this part of the resources, if part of a managed resource, between the two parts of how to achieve a better, from the customer perspective, it must be at least also co management of the resources, managed out of the two parts and multi tenant forms, not exclusive a state.


Hangzhou also beat cloud CTO Huang Huipan: what is the cloud acceleration

Hangzhou and pat cloud CTO Huang Huipan delivered a speech entitled "what is cloud acceleration" speech. He first introduced in the end what is cloud acceleration, he believes that the nature of the network is accelerated, after which he also shared the UPYUN cloud computing system architecture. Although cloud acceleration will bring the cost of hardware, system development and operational aspects of the network cost, but cloud acceleration brings the value is obvious, he thinks there are two main reasons, first is the service standard, transparent, fair; the second is service quality consistency and optimization cycle is short, and as a new generation of CDN Accelerated Service UPYUN redefined the cloud acceleration.

Here said the cloud acceleration refers to the hardware resources are available at all, they are just a fine sand throughout the network, there is no state. It will be a unified component of the standard to be combined to provide the same service. And then any component of the damage it will not affect the overall quality of service. This is the real cloud. If any server failure will affect the quality of the service, it is certainly not a cloud. Cloud acceleration it is like this, and then cloud acceleration in the end is what things, CDN this decade has been in contact with a lot, and then the traditional CDN which will sell what? There are static acceleration, dynamic acceleration, there are small files to accelerate, there is a large file to accelerate, as well as video acceleration, as well as direct speed, a lot of a lot of. In this inside, it is divided into many types of acceleration, respectively, a lot of packaging products. , of course, in their company, even there is a so many products specialized network of services, such as large file service is a network, the small file service is a network, the dynamic request of acceleration is also a network.

UPYUN system architecture, the only use of different technology in foreign countries is to use IP advertising model to do intelligent mobilization, the domestic no way to do, the cost is too high, so the domestic basically use intelligent DNS to do scheduling. To the nodes inside, in fact, are the same, the general Internet Co will use LBS to do load balancing, each node 15 servers, a large 30 servers. But this inside, although the structure is the same as the traditional, but in fact the inside of the thing is completely different. Inside a node, a cabinet server, all of the hardware, including the inside of the software are the same. The only difference is the name of the machine is not the same, IP is different is that the hardware is known, a server or a node, just the whole cloud accelerated network inside which a small unit, a small sand can readily damaged, cloud acceleration system, all are each other, including to the switch, all hot standby, can always suspected in any one device off.


The data members of the Committee of experts, Shanghai Sheng Tao Big Data Technology Co., Ltd. General Manager Ding: big data intelligent IT tools and platform

China, member of the Committee of experts on big data, Shanghai Sheng Tao Big Data Technology Co., Ltd. General Manager Ding in the theme is "big data intelligent IT tools and platform" speech first examples of large database market exists many problems, (1) inheritance -- does not support acid, big data will always be "from the sideline while talking"; (2) performance, high data not only "big", but also to "fast". High performance, real time big data is the future of the road; (3) easy to use - big data to bring us not only the "value big", more important is that everyone will use the big data.

He introduced to their system, the core idea is the whole system is built with a series of building blocks and method, and users can develop what they want and building blocks can be become the building blocks of your own, can protect your own business interests. And the whole system is easy to use, and can give us a lot of security.


More exciting content, please pay attention toThe seventh China Cloud Computing Conference live coverageSina Weibo, Sina Weibo@CSDN cloud computing.

top
Zero
step on
Zero