研究目的
To present an adaptive framework to cope with the problem of fault tolerance in cloud computing environments by employing both replication and checkpointing methods to ensure cloud reliability and availability.
研究成果
The proposed adaptive framework improves the performance of the cloud in terms of throughput, overheads, monetary cost, and availability by dynamically selecting the appropriate fault tolerance method (replication or checkpointing) based on the current conditions of the cloud. The framework's adaptive nature in determining the number of replicas and checkpoint intervals gives it superiority over existing methods.
研究不足
The framework's performance is evaluated through simulation, which may not fully capture the complexities of real-world cloud environments. Additionally, the framework assumes that the failure probability of virtual machines follows a Poisson distribution, which may not always be the case in practice.
1：Experimental Design and Method Selection:
The framework employs both replication and checkpointing methods for fault tolerance. It dynamically selects the suitable method according to the current conditions of the cloud.
2：Sample Selection and Data Sources:
The cloud used in experiments is generated with 100 heterogeneous virtual machines connected with fast Ethernet technology (100Mb/s). The number of data centers used ranges from 5 to 10, each containing 4 hosts.
3：List of Experimental Equipment and Materials:
Each host’s memory is 10 GB, storage is 2TB, and processing capacity ranges from 1000 to 10000 MIPS. Each virtual machine has a memory of 4 GB and one computational unit.
4：Experimental Procedures and Operational Workflow:
The framework's performance is evaluated in terms of throughput, overheads, monetary cost, and availability through simulation experiments with a variable number of customers’ requests (from 500 up to 2500 requests).
5：Data Analysis Methods:
The performance of the framework is compared with existing algorithms (OCI algorithm and a static replication-based algorithm) using metrics such as throughput, overheads, monetary waste, and availability.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
