cloud multimedia applications, service and devices, multimedia delivery is expected to become the major traffic of Internet which will keep increasing rapidly. In order to serve such large scale multimedia applications, more and more service providers store their video assets in the cloud and delivery streaming to their consumers cross cloud, for example, YouTube. Along with the growth of users and the amount of media content constantly being produced, traditional cloud-based storage has two drawbacks. First, a lot of servers and storages devices are needed, which could easily be the performance bottleneck in the whole system. Second, to provide differential classes of services in the large-scale situation, system tends to need many additional devices.
This article proposes a robust, scalable, highly available and service level provisioning cloud-based storage system designed specifically for distributing multimedia content. The proposed system contains a proven Adaptive Quality of Service (AQoS) algorithm in order to provide differential service levels. The system can also be used flexibly in large, medium and small-scale environment. In addition, some algorithms are also developed to increase overall system performance and fault tolerance. Implementations and experiment results show that the proposed system can meet the requirements both in the laboratory and a practical commercial service environment.Quality of Service Provision in Cloud-based Storage System for Multimedia Delivery
Previous work on storage resource management can be classified into two classes largely. One is, guaranteeing each client’s storage QoS requirements set by a system administrator. These systems such as the required response time objectives by regulating the rate of other clients’ workload incoming into the storage system uses Earliest Deadline First (EDF) to meet the response time objectives but it is impossible when an unexpected workload burst occurs by other clients.
Chameleon uses leaky-bucket with feedback control but leaky-bucket system does not use the storage system efficiently because it is also not work-conserving. Triage adopts a control theory to predict the system performance and correspondingly adjust its system model for performance isolation and differentiation. Its system model is not sensitive to the performance dynamics perceived by concurrent clients due to different physical data position.
We proposed system contains a proven Adaptive Quality of Service (AQoS) algorithm in order to provide differential service levels. The system can also be used flexibly in large, medium and small-scale environment. In addition, some algorithms are also developed to increase overall system performance and fault tolerance. Implementations and experiment results show that the proposed system can meet the requirements both in the laboratory and a practical commercial service environment.
We design and deploy the storage system with QoS provision for a multiple class-aware multimedia delivery service. The design goals and requirements for a storage system should include that the system can be deployed in cloud and also can be used flexibly in large, medium and small size environments and has features of scalability, considerable fault tolerance, security and other basic requirements.
Our proposed system applies the algorithm on the targeted storage which has been decided in the previous step. This algorithm is applied on the targeted storage with its own statistical data. each storage has its own statistical data for calculation. These constants are used in the algorithm are set in the configuration file and can be changed at runtime. Storage I/O benchmark tools (iozone) are used to perform the R/W pattern which is close to our services to decide the constant values.
Since storage I/O throughput is very dynamic and the system only needs rough values. If new storage is the same as the storage used now, the constant value could adopt from the current value. If size is the only difference between the new and current storages, modification of some constants is needed to fit into the new storage size.