Font Size: a A A

Research On The Elastic Service Management Framework For Cloud Environments

Posted on:2014-08-10Degree:MasterType:Thesis
Country:ChinaCandidate:B XiaoFull Text:PDF
GTID:2268330422962500Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
With the advent of cloud computing era, cloud platforms are gaining popularityamongst developers and drawing attentions to them. However, currently hosting complexdistributed applications in the cloud still lacks mature solutions. There are some key issuesmust be solved to help developing and deploying SOA applications onto IaaS clouds,which involves efficiently implementing highly available and elastic service-clustermanagement mechanism, ensuring service availability when IaaS infrastructure scales, aswell as providing systematic supports for customizable load-balancing and fault-tolerantstrategies in service implementations.Elastic service management framework for cloud environments solves theses issues. Itcomprises a service-cluster management solution and a cluster-aware RPC mechanism.The service-cluster management solution provides SOA applications with a dynamicscalable runtime architecture and enables them to be aware of system topology changes inreal-time. It also allows dynamically adding or removing service instances to or fromservice-clusters at runtime and notifying service callers in real-time without interruptingservices, and enables each system component to quickly and accurately access systemtopology information with consistency. Cluster-aware RPC mechanism added servicediscovery, dynamic configuration, load balancing, failure detection and fault-tolerance toRPC system, which hides the implementation details of underlying system, simplifying theprogramming of distributed services on cloud platforms. Via these two core functionalities,the framework provides a solution for the developing, management and monitoring ofhighly available, and elastic applications on IaaS platforms.Functional tests show that service-cluster management solution achieves real-timetracing of system topology and status of service instances, the cluster-aware RPCmechanism also realizes service discover, load-balancing, fault-tolerance as expected.Performance tests indicate that service-cluster management system deployed on EC2withT1.Micro instances can support SOA applications with up to2000service instances.
Keywords/Search Tags:Cloud computing, Service management, Elasticity, Fault-tolerant
PDF Full Text Request
Related items