May 23-May 25, 2012 cloud base of China Cloud Computing Conference opened the curtain, show cloud practice, explore cloud trend, share cloud future. The cloud base shows the latest technologies and new products in the field of cloud computing, and discusses the development trend of cloud computing, the great value of large data, the energy-saving of data center in cloud times and so on. Tian Yun Technology Skyform Research and development team director Zhang Yi introduced days and so friends characteristics and design ideas, and share cases.
Zhang Yi: Everybody good afternoon, I divide two parts, the first part introduces the characteristics of the sky platform and our design ideas, the second part share two cases with you, our cloud platform in the actual use of customers will be used for what purpose, what kind of services.
Skyform Market Positioning just now Zhang always mentioned that we are an open and neutral principle, for the cloud platform, we are only referee, not players, we are open neutral cloud management platform for virtualization technology, we do not have special technology lock-in, the current virtualization technology, including X86, Some of the physical machine virtualization, such as KVM and so on, this series we do not have a special technical lock, for the heterogeneous hardware we do not have a special preference. Therefore, the entire platform should be a neutral platform, it will provide services and operation of two functions.
Cloud Technology Skyform Research and development team director Zhang Yi
There is also a topic we want to think about, mainly skyform cloud platform in the end is resource-oriented or service-oriented, how to understand it? Service-oriented we say is driven by the real needs of users and businesses, our main concern is the entire service, rather than a single resource, and ultimately by the top of the service to determine our demand for the underlying resources, the entire platform is a top-down design is a design way, In turn we can also see that there is certainly a top-down design, we are focused on resource management deployment monitoring, we focus on the resources themselves, ultimately led to the resources to determine the upper service, this is what kind of concept? For example, I give my users to push our cloud platform, our resources management How to perfect, how good, finally said you can provide what kind of service, I said I can support your business, the customer again said my business is so so, how do you support it? I now turn to see, my resources management may be away from the service real business will still have a certain gap, I will tell customers you these applications I can support, other applications you need to modify, temporarily can not run on my cloud platform. Then we find it strange, why? Do users buy your cloud platform to change my business? So, we have always advocated that we should be oriented to business, driven by real business. Therefore, we in the whole platform design, first of all, from the business requirements, analyze the user's business requirements, the business to construct the service template, because what kind of service template to support what kind of business. So we are concerned with the process and service patterns of the entire service.
is service-oriented, and then encapsulates the resources that our business needs, and of course it falls to the resources or the computing, storage, networking, software, and so on, and ultimately the underlying resource structure determined by the top services. We summarize the architectural characteristics of Skyform, is a top-down, to serve as the main line throughout the entire product, service is from the user and the actual needs of the business to start, through the service of unified organization and perspective of some resources, this is characteristic one. Feature two, resource pool management is only a means, not an end. What is the purpose? The aim is to support the service, to support the business, so the business is to decide the service, and then the service decides the resources.
The third feature is that our underlying resources and resource management adopt an open and cooperative strategy to achieve efficient integration and management of resource pools, a task that must be done.
This is the Skyform overall architecture diagram, this is a logic schematic, the bottom or a series of resource pools, including our computing storage and network, in our resource pool management, there will be two more important one module, is the resource template management and a resource instance management. In our top has the Operation management platform, the Operation management platform has two main modules, corresponds with the resource management is the service template management and the service instance template. What do you call a service template? The service template is defined by your upper support business, the service specification is a kind of, is by the service template to manage. For example, service template analysis of our application, perhaps the most typical is the application of Java EE, may be more typical of the three-tier architecture, the three-tier architecture itself is described as a service description, service template and resource template is what relationship? It should be a nested relationship, the service template defines what resources I need, and the resource template defines how the resources in my resource pool are normalized, creating a resource template, and how these resource templates can be combined into a service template to run, which is a static definition. From a dynamic point of view, with the service template you can create a service instance through this template, such as our three-tier architecture requires a set of server resources, a set of storage resources and a set of network resources, which are opened as an instance of life cycle, from the opening of resources, the use of the end of recycling, The entire service pool management is placed on the operational management platform, the service instance can be disassembled into many resource instances, the so-called resource instance is just a single resource specification defined in the resource template, which is managed by resource management to handle the lifecycle management of individual resource instances in each resource pool, and, correspondingly, with a portal above us , the portal may be based on the use of different objects, there will be the end user portal, the Operation management portal, as well as the resource management portal, the corresponding we will have a cloud platform monitoring, including equipment monitoring, and so on, all of these components constitute a platform for us.
There is also a more important module is not said, is our resource scheduling. The so-called our resource scheduling is to look at our service requirements and resource management matching. How do you say that? Our resource pool has a lot of resources, resource pool can also be very large, we serve the application of the example, how to go to specific resources, this is the need for resource scheduling to make a decision. What kind of service example should be opened in which more appropriate resources, this is a decision, the decision-making body is resource scheduling, it can make our huge pool of resources to work methodically and efficiently.
This is a more detailed function of the display diagram, we just said that these layers of service modules will have some small detail some of the modules to support.
Now look at the kind of service templates that this product can provide, from the cloud hosting services, we can provide small machine services, X86 physical machines and X86 virtual machines, from instance PAAs we can support two categories, one is the Java-EE three-tier architecture of the service model, as well as the cluster architecture pattern, Basically these two categories currently cover most of the applications in our industry.
also has a security service, like we set up a security service, virtual firewall services, as well as cloud monitoring, cloud monitoring services are mainly used to provide customers with his own application of cloud resources monitoring services to use, as well as cloud storage, including our typical block storage devices, as well as object storage devices, file storage, There is also a cloud network service and a cloud backup service. All of these service templates can be used individually or in combination to meet the needs of the business system that supports our current customers.
complete self-service process, the user can through a similar way of shopping, can apply for approval, the ultimate use of resources, this is a self service process.
for the management of heterogeneous resources in multiple resource pools, we have just mentioned that from hardware, we support many manufacturers of different brands, different architectures of these physical devices, from the virtualization platform, we also support a lot of virtualization technology. From the Third-party interface, we can now support Amazon, EC2 interface.
below quickly to share two cases. The first thing I want to introduce is the operator's private cloud, the general background of the project, there are many operators idle some resources or some new resources, including a lot of X86 physical machines, and minicomputer, storage is also a lot, our goal is to achieve resources on demand on these resources, dynamic scheduling and monitoring , this is demand one. There is also a part of the demand that requires our users to apply in a self-help manner, not as they have previously procured or assigned by a fixed person. The third requirement is an efficiency requirement that requires our IT resources to respond efficiently and with high service capabilities, which is a general introduction to the project.
This also encountered some problems, such as the physical machine to use the model, there is no way for users to flexibly configure resources, its scalability and high availability is insufficient, this is to solve a problem. There is also a problem is X86 server and small machine, the two types of devices are co-exist, for heterogeneous resources, basically difficult to unified management. Therefore, the management is complex, the operation dimension cost is comparatively high.
Another challenge is that the previous approach is to manually implement the device application, approval deployment to the configuration, the final resource-ready cycle time is very long, the responsiveness of the business is weaker. In response to these several challenges, our cloud's solution is to use our unified virtualization technology to integrate resources, X86 servers and minicomputer We all will adopt this virtualization technology, in this case X86 server virtualization is basically using Citrix, we use the Solaris technology to do virtualization. Compute resources and virtual servers to bind, the computational resources you require are not directly related to a specific physical device, you want a capability, not a real machine, and we can increase the utilization of the server and reduce the input of it hardware through virtualization and resource sharing.
for the unified control of heterogeneous resource pools, we provide a unified management view, so, in our one resource platform monitoring can not only do to the X86 physical machine management, but also to the small Computer Management, the administrator can use the same way for these two types of management, reduce the complexity of management.
We also provide users through the cloud Platform's portal system, complete IT resources from approval to deployment, to use, to recycling, the entire service process is self-service automated, we have simplified the previous IT resource allocation management of some complex operational processes, improve the resources of the business responsiveness. Here, we support multipole approvals, this is based on the actual needs of a customer, although we are doing an automated process, but also to complete the way the previous artificial paper to provide a multipolar approval, there is an allocation of IP resources, the original resource management platform is basically rely on dynamic IP to allocate IP resources, This customer also needs us to be able to enter the IP address of the resource, which we can also support.
The deployment of the entire program, in the following about dozens of X86 resource pool, the cluster of physical servers, there is a Fujitsu small machine, it runs the Solaris Container, which we have with the small computer resource pool has an adapter, can do small machine resource management, X86 Resource management can manage X86 resources, in the above with the operation of the business related to the approval, user management, statistical reports, system monitoring and so on, for resources, we will have a resource perspective, as well as resource monitoring, we will provide the ultimate User a portal, to our own system administrator to provide a portal, At the same time for the end user to provide an operation of the virtual machine Mobile panel, it can be through our portals to declare approval, and ultimately obtain these two types of resources, and then use the VM, this is our final case. What is the end customer gain? He can manage heterogeneous resources in a unified, improved resource utilization and management efficiency, through automated management, we can reduce its maintenance costs. Through resource consolidation and sharing, IT resources are saved, and ultimately we optimize IT infrastructure, build a service-led infrastructure, and improve IT service capabilities.
This is a case of the private cloud within the carrier that we are pushing in the operator. Another application scenario is in the park, now Software Park, or industrial park, the park is basically doing cloud services, we do a case is the Shanghai Pudong Software Park, Software Park, the name of the project online IT Services cloud, The main goal of this project is to provide a computing environment for the small and medium-sized science and technology enterprises in the software park, and reduce the hardware investment of the enterprise it, you can do a bag to stay such an effect. By providing enterprise management software, it can also reduce the management cost of some enterprises, save it investment, and develop investment, attract more enterprises to stay in the software park. We also offer PAAs and SaaS platforms, and companies are making a transition to cloud computing.
This is the framework of the entire remit-wisdom online system, here is also the use of our cloud platform to build, but also server virtualization, network virtualization, storage virtualization, the specific resource pool management, also divided into two parts, the following part is actually using the resource management platform we just mentioned, Included in this project is called the operational support subsystem, probably the same function, support resource management, automated management, monitoring, configuration, plus some other process services, backup and so on, plus these features, which constitute the entire operational support system, there is a corresponding business support system, This can correspond to the operational management platform of the product, some of the main modules it provides are customer relationship management, billing, order, billing, billing, etc., which will provide a portal on the left, a user-service portal, a portal for ISV management, and, on the right, a service management subsystem that involves service design , service development testing, this is the definition of the service template, after the service template is fixed, there is a service release, after the service is released, will enter into the service directory, the user can finally see a service directory, from its service portal can be ordered in the service directory to provide all of these services.
just now we have been referring to services, in the remit of wisdom online projects are mainly to provide such a few services, cloud host services, resilient block storage services, backup services, including virtual firewall services, as well as virtual load balancing services. In this case, we also provide SaaS access services, not SaaS services, but SaaS access, and we do not develop specific SaaS applications, which are provided by third parties, Our platform has the ability to access these applications from third party SaaS to our platform for unified management, unified billing, unified monitoring, and providing a SaaS service to our final users.
End Customer benefits do so, integrate High-performance servers, as well as high reliability storage devices, as well as high-quality network bandwidth, the provision of IT resources immediately rented immediately, and the billing model is very flexible, which is its first benefit. Provide flexible supply to users, quickly deploy IT resources, mask some of the complexity of the infrastructure, and ultimately simplify some of the customer's operational management costs. Fully meet all types of enterprise users, as well as individual users of it cost reduction, but also high reliability, easy to manage some of the needs.
Thank you.
(Responsible editor: The good of the Legacy)
