Research On Intermediate Data Fault-Tolerant Approach In Cloud

Cloud computing is considered the development of distributed processing, parallel processing and grid computing, or implementation of business. With the rise of cloud computing applications, its scope may be further expanded, the processor there is also increasing the probability of error, fault-tolerance has become a major research of cloud computing, and intermediate data fault-tolerant is an important component of fault-tolerant. Intermediate data loss seriously affects the computation of the task. Therefore, the intermediate data fault-tolerant technology research and implementation becomes an important research topic in the field of cloud computing..In this paper, according to our event detecting research on intermediate data fault-tolerant, we propose a layered approach for intermediate data fault-tolerant named LIMF, specific working is as follows.In the analysis of the existing cloud computing architecture, this paper gives a cloud computing architecture named CAIMF (Cloud Architecture for Intermediate Data Fault-Tolerant), according to different nodes have different functions, detailed description the composition of datacenter nodes.Methods for the single function of the cloud processing operations limitations. Aiming at the issue that a job can be split into different tasks, this paper analyzes the data dependency, and gives different detailed approach between task and task, also gives detailed approach within a task.This paper propose a layered approach named LIMF, and propose intermediate data select approach, fault-tolerant data placement approach and fault-tolerant data create approach to support intermediate data fault-tolerant, put up detailed designs and give relevant approach for them.We design two fault-tolerant data sheet used to record data and give the filling method, and after the failure of the machine give two scheduling methods.Finally, java language is used to do some experiments for verifying the feasibility of a layered approach for intermediate data fault-tolerant. When has a failure, the approach which is proposed in this paper effectively improves the computation time of application, reduces the load on the main server.