Font Size: a A A

Optimal Deployment Strategy And Service Migration Based On Staging And Business Predicting

Posted on:2015-01-11Degree:MasterType:Thesis
Country:ChinaCandidate:Z Y DiFull Text:PDF
GTID:2268330431454674Subject:Computer technology
Abstract/Summary:PDF Full Text Request
With the rapid development of Internet technology, in order to reduce operating costs and improve operational efficiency, the traditional enterprise customers can enjoy the service of software, hardware and maintenance and the right to use software and upgrade through internet by only spending some rental fees, rather than spending a lot of money on hardware, software, technician and maintenance as before.SaaS platform uses multi-tenant mode. However, for the service delivery system of multi-tenant, tenants demand service providers to meet their performance and other QoS requirements of services, such as, whether the allocated resources are adequacy, whether the response time is satisfy with the tenants and so on. But, the services providers hope the SaaS platform should have the ability to take full advantage of the resources. Thus, the tenants want to hire high performance services and services providers demand minimizing cost at the same time of meeting tenants requirements. But they are often contradictory between high performance and low cost. Therefore, in order to balance the contradiction, this paper proposes a two-stage optimization strategy based on the initial service deployment and runtime.This paper develops the Benchmark based on the supplier management system(S-BM). When tenants applications are the first time to deploy optimally, we use the method of iterative test continuously based on S-BM and staging. And for the second optimal deployment of runtime, this article proposes the solution that combine real-time load measurement model with key business predicting based on Markov chain to make migration strategy, synthesized the migration cost. We also design and implement these two methods and analyze their effectiveness through experiments. Finally, conclusion and prospect about our research work are provided in this paper.
Keywords/Search Tags:Multi-tenant, Optimal Deployment, Staging, Business Predicting, ServiceMigration
PDF Full Text Request
Related items