Cloud computing Drives data center Operations Service Innovation

December 22, 2011, with the support of Industry informatization Department, Information Propulsion Division, China Data Center Industry Alliance, China IDC Industry Alliance Network undertook 2011 China Data Center construction and Operation Dimension High-level Forum opening, the Forum with "leading technology (000669, stock bar) to create the future" as the theme, The meeting will be a new generation of data center planning, construction, operation, and security and other substantive issues of extensive discussion and interactive exchanges. Room 360 to this high-level forum to do a full broadcast. The following is a full transcript of Fan Chunying's speech:

Fan Chunying: Dear friends, everyone it Tongmin good morning. I am from China million network Fan Chunying, today is very happy to have the opportunity to share with you in the cloud computing and data center in the field of discussion. The topic of my speech is "cloud computing drives data center Operation service Innovation". Million net everybody also very familiar with, this year is our 15 anniversary, million net also accompanies a from scratch process, from 05 million nets joined Ali also make million nets have a new development opportunities, to today we are in the transformation of all IDC and all the construction, began to promote the cloud computing technology as the core of the new IDC construction and practice.

We see that in the application of the Internet customers have many often encounter pain, such as a single point of failure, upgrade, very poor expansion, very difficult, there are users of the initial construction cost is higher, we are commonly known as low price to enter the threshold high. When a user is purchasing a server, or building his own data center, he may need only a fraction of the resources, but there is no way to provide them with resources alone, forcing them to purchase a large amount of idle resources. For example, we found that in IDC storage resources and computing resources in fact have to get 30% of the use of the other 70% is in the form of hardware in that idle, which causes a lot of waste and cost input. And for Internet operators, will also face similar problems, such as they will encounter tens of thousands of server management problems, tens of thousands of server inventory can not be reduced, but we can not go to a more out of the resources released to other customers, we can only continue to build new data centers, in the continuous adoption, But the overall is not low carbon and green, and deferred delivery of things have occurred.

Therefore, in this case cloud computing will play its role came into being. When we find a simple storage hard disk can not meet the time, we use an array to meet, if we can not solve the cloud computing, distributed computing to meet. When a server is not able to support, we think of the cluster, with multiple servers. But when the machine also has bottlenecks, we can only use distributed computing to solve this problem. The network is the same, and finally we need to link multiple data centers together, to form a cloud computing data center, which is integrated together, we can integrate three kinds of computing storage and network capabilities, and can output separately, so that the value of cloud computing will be realized, this is the cloud computing service capability output.

So we can see the core value of cloud computing to data center, the first is to reduce operating cost, improve business and management efficiency, second, improve the reliability and carrying capacity of it application, third, the cluster can achieve scale effect, so that the overall TCU can be reduced. Here is a display of cloud computing market size, and IDC market size. China's IDC scale is basically less than 20% a year, we can see a table like the lower left-hand corner of about 10%, and cloud computing has always been by leaps and bounds, especially in China, to maintain a very high growth rate.

As we can see in this picture, this is a business model map of cloud computing. Generally we will be divided into Iaas,saas,paas. and IDC Data Center layer is basically in IaaS this level, million net is also at this level. At the PAAs level, we will see many well-known companies, and internet leaders, such as Google, Baidu, Salesforce, such as the architect. But there will be new open platform competitors in this area, such as the mobile internet, Sina Weibo, 360, an open platform that integrates Internet applications and mobile Internet applications into an open platform, and is able to centralize resources to provide to individual ISVs.

In the area of SaaS, we'll see a lot of service providers that specialize in enterprise applications, including SAP, 800, and Amazon, which provide standardized services at the SaaS level. In the image above is the terminal user, as you can see here, like SAP, when he went to service in the SaaS field, he didn't actually do all the PAAs and IaaS things, so he bought a company where the middle PAAs could see a chart, and as an intermediate PAAs tier service provider , in fact, a server is not, all servers are rented EC2 services. So, this can form a very good upstream and downstream relationship based on the cloud computing biosphere, from SaaS to PAAs, to IaaS.

To share with you, this is in 09 before the Million network IDC architecture. There are a large number of computer rooms on China's map, there are about 4 rooms in Beijing, Dalian Netcom has room, Shanghai, and Guangdong, in fact, there are some not written on the top, very fragmented. Some of the aims are for localization services, some for solving the problem of interconnection in China. In short, we have to continue to open up a lot of small room. In 2000, these things have changed, since 2010 million network IDC began to tilt to the cloud computing, we began to merge three of the most core room, respectively, in Beijing, Qingdao and Hangzhou, the room between the several of us through the bare fiber connected together. At the same time, we will deploy a lot of CDMA nodes on each other single point, to solve the problem of interoperability, so we have formed in Beijing, Hangzhou, especially the two major nodes as the core of the cloud computing center, where they can be remote disaster recovery backup, as well as some resources to share, so you can put a lot of scattered room inside the server , and tilt the focus of these large cloud computing centers. If the user is a hope to be able to full network coverage, we can put the core business, and his original station on the Internet in the Internet center, and to reach the various provinces and cities Terminal Services placed in each CDMA node, which formed a very benign alley network. At the same time, we also save a lot of 7x24 management costs.

Here to share with you before 09 million network IDC operation and maintenance platform has a variety of systems, here show 6 systems, including black box, white box monitoring, as well as traffic monitoring and so on a variety of. In fact, the complexity of management is very high. So, when it comes to cloud computing, the data center transforms, we also do some integration of each system. Mainly divided into Cloud computing center management system, and intelligent monitoring management system, and we integrate virtualization technology and distributed storage into which, through a set of cloud computing system to manage all the resources and monitoring.

The following is what you see here is our main cloud host system architecture, the following is the data center we write will have three data centers between each other fiber communication. In the data center is a layer of resources, for example, computing, storage and network, and then up one layer is our management system management platform, management platform is our virtual machine cutting all virtualization, in the UP CMP, cloud computing management platform, as well as billing system and BC users to do the purchase and display, CMP is the control system that customers can see, filing system and so on, these are built on the cloud host. In this way we will be formed regardless of our user support, engineer support, can be in the same set of systems to complete all the work, the future will continue to work in this direction.

After the cloud computing data Center is built, we can provide more services to users. For example, traditional IDC we are only complex hosting, there are some monitoring management. But in fact, the higher value-added work is not reflected in it, we now hope to be able to use cloud computing to bring more value-added services to enterprises, such as we can provide a complete cloud computing solutions, we have a multi-line room, some customers with virtual machines can not solve the problem, we could take the virtual machine as a front-end To continue to implement the database through physical machine, the virtual machine and the physical computer intranet interconnection, you can reduce the user to move to the threshold of cloud computing. At the same time, we provide customers with more consultancy services, as well as flexible payment mode. For example, we can now support a monthly payment based on the amount of resources used. Furthermore, we provide more 7x24 VIP services, as well as security scans, system monitoring, and regulatory and filing support.

Below we see is now cloud host can realize some traditional physical server cannot implement some new features, for example, this diagram shows the storage, in the distributed storage, we can implement a data save three, and belong to three hard disk, three different network. Under a switch there will be a hard disk to correspond, so that at least to ensure that the same data, at the same time redundancy saved three copies, will not be due to a hard disk missing caused data loss. And we support the fast, do a fast every day, every 7 days, a customer 1 months will save 10 backup, when there is a problem, the data mistakenly deleted, or hacked and so on these things, he can go back to a certain point in history.

This is also the advantage of distributed storage technology, if the traditional way to do backup, you need to copy the data, the copy process will be full, and affect the user's operational efficiency. In distributed storage, we don't need to copy files, we just need to be like a dot roster, every day to point down, this is very good technology, for example, we have 200 people in the venue in accordance with the whereabouts of 200 people, tomorrow there will be added to the note, each of the quick tune can be completed within a few minutes. So, give the user a quick tune every day without affecting his use. If the customer needs a permanent quick transfer, we can save one of the quick calls, will not be deleted over time, or the phenomenon has expired.

Another new feature is the failover mechanism, which is what virtualization technology brings to us. When we can contribute to the storage of the sun, when a server fails because he is storing it in another storage share, the system discovers the failure, then restarts the virtual machine on the other available server, and automatically maps the backward data to the new health virtual machine, so that the failover can be realized.

Dynamic upgrades are also the same concept, we can for a small virtual machine, as the user's growth needs to upgrade to a larger CPU, larger hard disk space, and bandwidth. Here, in addition to memory, other Ziyun can be completed, more than the traditional physical server to a more advantageous place. At present, our solutions for different stages of development are mainly cloud host, cloud storage, this is the basic model. The second mode is cloudengine+rds, this is our cooperation with Aliyun, for example, there are 10 virtual machines, if in accordance with the general mode can only serve 10 customers to function alone. But we can connect 10 virtual machines together, available to 100 customers, we can divide 100 resources into 10 virtual machines to provide to customers, when customers grow very well, when he needs to take up a virtual machine, can be upgraded to a virtual machine, or he can become more than one virtual machine, This makes it more flexible to pay for resources and a lower starting point. RDS is similar to Amazon providing database services that can currently support distributed databases. So we can use one of the largest separate database clusters, which is also an inevitable choice.

Mixed cloud solution, if a customer's demand is already greater than the first independent physical machine support scope, there is no need to do the virtual machine, we need to put those physical machines and virtual machines can be mixed, so we do not simply require customers to migrate virtual machines, the key according to his needs, We can give him more cost-effective solutions. Below see is million network now cloud computing platform on some typical customers, such as see the mobile internet and electric business customers have a lot of, such as Taobao, now Taobao independent software developers, now more than 50% is running in the million network cloud computing platform, including the recent double 11, double 12, Many big promotions are million network cloud computing platform in the provision of core marketing support, including group buying category, as well as invoicing ERP pointed out, but also organic front nets, Amoy and so on to do the emerging internet industry, including Mushroom Street.

So, we are also exploring the integration of the mobile Internet with cloud computing in E-commerce. This is the traditional industry we have a lot of cases, now Haier, all global web site projects, Haier COM is running in the million-net cloud computing platform, as well as Schneider Electric in China Collection, intelligent control design is also operating under the million network platform. Here are a variety of ISVs that are the most expansive developer markets in the future. In here, we also do a lot of support for ISVs work, now Taobao buyers ISV, so now are free support, when half a year later can grow up later to charge. Therefore, through a relatively deep mode of cooperation, from the very fragile, very embryonic stage began to support these developers, hoping that they will be on the open platform to do more strong, and gradually grow up.

Finally, we hope that million network can become your partners and business success behind the brilliance, I hope you can start from million nets, thank you.

(Responsible editor: Lu Guang)