QOS-AWARE DATA REPLICATION FOR DATA-INTENSIVE APPLICATIONS IN CLOUD COMPUTING SYSTEMS
Cloud computing provides scalable computing and storage resources. More and more data-intensive applications are developed in this computing environment. Different applications have different quality-of-service (QoS) requirements. To continuously support the QoS requirement of an application after data corruption, we propose two QoS-aware data replication (QADR) algorithms in cloud computing systems. The first algorithm adopts the intuitive idea of high-QoS first-replication (HQFR) to perform data replication. However, this greedy algorithm cannot minimize the data replication cost and the number of QoS-violated data replicas. To achieve these two minimum objectives, the second algorithm transforms the QADR problem into the well-known minimum-cost maximum-flow (MCMF) problem. By applying the existing MCMF algorithm to solve the QADR problem, the second algorithm can produce the optimal solution to the QADR problem in polynomial time, but it takes more computational time than the first algorithm. Moreover, it is known that a cloud computing system usually has a large number of nodes. We also propose node combination techniques to reduce the possibly large data replication time. Finally, simulation experiments are performed to demonstrate the effectiveness of the proposed algorithms in the data replication and recovery.
Due to a large number of nodes in the cloud computing system, the probability of hardware failures is nontrivial based on the statistical analysis of hardware failures. Some hardware failures will damage the disk data of nodes. As a result, the running data-intensive applications may not read data from disks successfully. To tolerate the data corruption, the data replication technique is extensively adopted in the cloud computing system to provide high data availability. For example, the Amazon EC2 is a realistic heterogeneous cloud platform, which provides various infrastructure resource types to meet different user needs in the computing and storage resources. The cloud computing system has heterogeneous characteristics in nodes. Note that the QoS requirement of an application is defined from the aspect of the request information. For example, in, the response time of a data object access is defined as the QoS requirement of an application in the content distribution system.
DISADVANTAGES OF EXISTING SYSTEM:
vThe QoS requirement of an application is not taken into account in the data replication. When data corruption occurs, the QoS requirement of the application cannot be supported continuously.
vThe data of a high-QoS application may be replicated in a low-performance node (the node with slow communication and disk access latencies). Later, if data corruption occurs in the node running the high-QoS application, the data of the application will be retrieved from the low-performance node.
vSince the low-performance node has slow communication and disk access latencies, the QoS requirement of the high-QoS application may be violated.
We Propose QoS-aware data replication (QADR) problem for data-intensive applications in cloud computing systems. The QADR problem concerns how to efficiently consider the QoS requirements of applications in the data replication. This can significantly reduce the probability that the data corruption occurs before completing data replication. Due to limited replication space of a storage node, the data replicas of some applications may be stored in lower-performance nodes. This will result in some data replicas that cannot meet the QoS requirements of their corresponding applications. These data replicas are called the QoS-violated data replicas. The number of QoS-violated data replicas is expected to be as small as possible.
To solve the QADR problem, we first propose a greedy algorithm, called the high-QoS first-replication (HQFR) algorithm. In this algorithm, if application i has a higher QoS requirement, it will take precedence over other applications to perform data replication. However, the HQFR algorithm cannot achieve the above minimum objective. Basically, the optimal solution of the QADR problem can be obtained by formulating the problem as an integer linear programming (ILP) formulation. However, the ILP formulation involves complicated computation. To find the optimal solution of the QADR problem in an efficient manner, we propose a new algorithm to solve the QADR problem. In this algorithm, the QADR problem is transformed to the minimum-cost maximum-flow (MCMF) problem.
We propose a new algorithm to solve the QADR problem. In this algorithm, the QADR problem is transformed to the minimum-cost maximum-flow (MCMF) problem. Then, an existing MCMF algorithm is utilized to optimally solve the QADR problem in polynomial time. Compared to the HQFR algorithm, the optimal algorithm takes more computational time.
ADVANTAGES OF PROPOSED SYSTEM:
vWhile minimizing the data replication cost, the data replication can be completed quickly.
vWe use node combination techniques to suppress the computational time of the QADR problem without linear growth as increasing the number of nodes.
üProcessor - Pentium –IV
üSpeed - 1.1 Ghz
üRAM - 512 MB(min)
üHard Disk - 40 GB
üKey Board - Standard Windows Keyboard
üMouse - Two or Three Button Mouse
üMonitor - LCD/LED
- Operating system: Windows XP.
- Coding Language: C# .Net
- Data Base : SQL Server 2005
- Tool : VISUAL STUDIO 2008.
Jenn-Wei Lin, Chien-Hung Chen, and J. Morris Chang, “QOS-AWARE DATA REPLICATION FOR DATA-INTENSIVE APPLICATIONS IN CLOUD COMPUTING SYSTEMS” IEEE TRANSACTIONS ON CLOUD COMPUTING, VOL. 1, NO. 1, JUNE 2013