Research on Campus VOD System Based on Grid Technology
Abstract: An implementation framework of campus VOD system based on grid technology is proposed. The hardware composition and software hierarchy of the system are described. An adaptive scheduling algorithm for on-demand tasks is proposed.
Keywords: grid VOD campus VOD grid
At present, each university has established its own campus network platform. Due to the continuous updating and development of the campus network platform hardware facilities, the campus video on demand service VOD (Video On Demand) has also developed rapidly. The VOD system uses the advantages of network and video technology to provide on-demand and broadcast services for online news, teaching, information, entertainment and other programs for teachers and students, and provides live broadcast services for large meetings and events. However, there are still some specific technical problems, such as difficult to control the flow, unable to guarantee the quality of service, how to lay out the server cluster, how to share resources and so on. Grid technology is to connect various computer resources to achieve the true meaning of resource sharing. The application of grid technology can solve some problems in campus V0D system.
1 Grid technology application research Grid is a group of emerging technologies built on the Internet, is a network infrastructure of the information society. Combining high-speed Internet, high-performance computers, large databases, remote devices, etc., can provide more resources and achieve true resource sharing. Grid is an increasingly important research field at present, and is called the next generation Internet. With the deepening of grid computing research, people are increasingly discovering the importance of grid architecture, and a new grid architecture-Open Grid Services Architecture (OGSA) is proposed.
OGSA is an important standard of Global Grid forum 4 and a network architecture that is currently widely supported. OGSA is the design prototype of GT3, the focus is on combining Globus idea with Web service. In the OGSA framework, all resources (including computing resources, storage resources, bandwidth resources, software resources, data resources, etc.) can be abstracted as services. Using OGSA is beneficial to manage and use the grid through a unified standard interface. Web Service provides a service-based framework. However, Web Services generally face permanent services. In a grid application environment, a large number of services are temporary and temporary services, such as the execution of a computing task. Considering the specific characteristics of the grid environment, OGSA puts forward the concept of grid service (Grid Service) on the basis of the original Web Service service concept, which is used to solve service discovery, dynamic service creation, service life cycle management, etc. Service related issues.
2 Campus VOD system based on grid technology The service object of campus VOD system is all ordinary users who use campus network. Because there are many levels of users. Computers vary in level, so campus VOD generally uses browser-based Web viewing. The transmission of audio / video multimedia information on the network currently mainly uses streaming transmission schemes. This continuous time-based media using streaming technology in Internet / Intranet is called streaming media. However, in the existing on-demand transmission of most VOD systems, the video content still needs the entire end-to-end path from the server to the client. Because video streaming requires high transmission bandwidth, and the transmission time of a single video is very long, generally around 120min, this requires solving some technical problems: reasonable scheduling and coordination of multiple resources, etc .; some real-time requirements require Solve as soon as possible; the demand for ever-expanding video services also requires the unified management and scheduling of individual isolated VOD service systems geographically distributed on campus. The grid will break through the limitations of the traditional Web and connect all resources (including computing resources, storage resources, bandwidth resources, software resources, data resources, information resources, knowledge resources, etc.) in the campus network into a logical whole, just like a Like supercomputers, it provides users with integrated information and application services (computing, storage, access, etc.). In this virtual environment, the virtual organization will eventually achieve resource sharing and collaborative work, completely eliminating resource "islands." In this way, people use the resources on the grid as simple as electricity, providing a more reliable guarantee for the campus VOD system.
The guiding principle of the campus VOD grid solution is to adopt OGSA. For the data transmission between each grid server in the system platform, the internationally recognized Globus grid computing protocol conforming to the OGSA specification is adopted as the data transmission protocol. Globus is a platform that constitutes a grid infrastructure. Its role is equivalent to a grid operating system. It is responsible for managing grid resources, but it is still relatively far away from specific applications. If you write applications directly based on Globus, it will be as troublesome as writing windows programs directly using the underlying API. Therefore, there needs to be a bridge between Globus and the application, namely Cactus. With the support of Caetus, writing applications does not need to consider the specific problems of the grid, and even previous applications can be used without modification. Cactus is an integrated, general-purpose, open source computing science and engineering problem solving environment. Cactus allows users to convert programs originally developed on personal computers (regardless of whether they were originally written in C, C ++, Fortran, or Java) into parallel programs that can be run on virtual grid computers; Cactus provides a simple and abstract API call interface, shielding the complex characteristics of the system, thereby simplifying the user interface and obtaining a strong displaceability; Cactus can run on machines with different architectures, such as a single processor, Cluster computers and parallel computers of other architectures. So for those scattered around the campus. VOD servers using different applications can be used directly with the support of caetus.
The campus VOD system based on grid technology mainly uses the grid method to integrate multiple VOD services and other related information in the campus grid to provide users with a unified and transparent view. Users only need to choose their favorite videos, the system will automatically provide users with the best video resources for users to use. All of this is transparent to users. Users do not need to pay attention to which VOD host provides video services for them, nor do they need to pay attention to the complex technical implementation of the system background.
3 Key issues
3.1 System hardware composition The construction of the system is based on several resources of the grid VOD subsystem centered on each VOD server geographically distributed in the campus network, as shown in Figure 1. These grid subsystems differ in terms of network topology, software and hardware architecture, security, and authentication strategies. Various VOD servers are interconnected by campus broadband network using multiple routers and other network equipment, while some other computers in the grid subsystem serve as video servers and can be transparent to users. Each video server is attached to an independent large-capacity disk array, as well as a cache and control management unit. In addition, at least one grid management server connected to all VOD servers must be installed in the system to provide global dynamic and static Web Service service information related to management, such as online users, bandwidth occupancy rate, and bandwidth limit. , The address of the server where the video is located, etc. The information transmitted between the grid management server and the user and VOD server is only query and control information, but no streaming media data, and the bandwidth requirement is relatively low.
The grid management server is mainly responsible for the grid retrieval information service GIIS (Grid Index Informafion Service), specifically the indexing of comprehensive information such as the collation and security of some user information (broadcast materials) in the VOD grid. The video server mainly compresses and stores the media data, reads the programs on the memory according to the user's request, and transmits the software to the user terminal in real time. In addition, each video server also needs to start a number of information providers to be responsible for submitting some broadcasted information and the load status of each hardware resource to the grid resource information service GRIS (Grid Resource InformaTIon Service) in the local VOD server, in order Dynamically select video servers with lighter load and faster response to provide resources. The operation mechanism of the grid router is similar to that of the IP router, except that the grid router can both receive or forward resource location requests, and can also store the mapping relationship between physical services and virtual services. In order to prevent information flooding and routing loops, an upper limit can be set on the propagation distance of resource requests—forwarding hops. In addition, the number of resource matches can be limited to reduce the amount of communication between grid routers and improve the performance of resource positioning.
3.2 System software architecture The software architecture of the grid VOD system is mainly composed of three levels, as shown in Figure 2.
(1) Grid resource layer These resources include all databases, data warehouses and file directories in the VOD system, various storage devices, and various computing resources (including supercomputers and PCs), which are physically connected through the campus broadband network.
(2) The grid VOD service layer provides an integrated management and processing platform for the grid VOD system. Its main function is to coordinate the sharing of multiple resources and provide data storage, organizational management, distribution, retrieval, and processing certification for the comprehensive use of various resources. And other services.
(3) Grid VOD application layer Based on the VOD grid service layer, it provides a VOD integrated application environment oriented to the application field. For each specific application area, it provides Web services, user management, resource allocation, unified scheduling, and traffic monitoring. Provide usage policies and protocols, application software tools, and application development platforms based on the usage patterns and characteristics of resource information to establish a resource information processing and service integration environment.
The information of the entire system can be divided into two types of static information (such as network topology, network node IP address and software version, etc.) and dynamic real-time information (such as system load status), information sharing, especially dynamic information such as resources and load Sharing and management is one of the main features of the grid system, and it is also the basis for the smooth scheduling and operation of a grid system.
3.3 Scheduler In the entire grid VOD system, the scheduler at the application layer is the most complex part, and almost all the information maintained and managed in the system of the system is mostly the decision-making task of the scheduler. Whether it is a system-oriented scheduling strategy (such as Condoz-G) or an application-oriented scheduling strategy (such as Apples, Nimrod / G, Netsolve), their main scheduling objects are processors, memory, and memory for scientific computing tasks. The "bare" hardware resources such as the network, and the interface provided to the user are also mainly designed for batch processing. The scheduling target is relatively single, mainly for load balancing of the present node. When a user request for security authentication is submitted to the scheduler, the scheduler first performs preliminary filtering based on static information such as user IP, task resource requirements, and network configuration, and then collects detailed dynamic information. Select a video server according to a certain strategy to reserve resources, submit tasks and supervise the execution of tasks. During this time, the configuration of the file resources of the video server may be triggered, and the task will be cleared after the task is completed. Its scheduling strategy follows the following two principles:
(1) Local priority. When the local grid VOD subsystem is lightly loaded and stores required programs, it is directly scheduled locally without considering the global situation. The principle is based on two considerations: â‘ the delay of transmission across subnets is generally large; â‘¡ the bandwidth on the backbone network is generally limited, and the cost is also high.
(2) The VOD subsystem asks the principle of relative load balance. When the local load has reached a given balance point but has not yet reached the allowable upper limit, it begins to consider using the target planning method to select the appropriate subsystem in the entire network, so as to avoid doing when the load is very light Unnecessary global coordination avoids the passive situation that may be caused by inter-system scheduling when the local is fully loaded. When the local system load reaches the balance point, the scheduling management module starts to see if it is necessary to add a backup to the program.
Based on the research of grid concepts, related technologies, the exploration of grid applications, and the research and construction experience of campus VOD systems, this paper proposes an implementation scheme of campus VOD systems based on grid technology. The main feature is that it can be constructed On a variety of heterogeneous platforms, it has wide applicability and adaptability, supporting unprecedented scale and expansion capabilities.
Product application:
â– Products for commercial,household,lighting
Product features:
â– High strength,impact resistance and thermal stability
--Panel material using high quality polycarbonate
â– High resistance to deformation
---Long term use without obvious distortion
â– Flame retardant grade
----The flame retardant grade of panel is in line with UL94-V2
â– Adaptation property
----American Standard socket adapter all accord with NEMA
Standard
Colour: Black Brown White lvory Almond Gray Red
Wall Plate UL,Electrical Wall Plate UL,Wall Wall Plate UL,Double Duplex Outlet Wall Plate UL
Hoojet Electric Appliance Co.,Ltd , https://www.hoojetgfci.com