SRM基准测试中工作负载定制与性能评估
|
摘要
为解决多租用者应用程序所面临的经营规模和运行特性等最佳配置策略问题,创建了名为S-BM基准的模拟供应商关系管理应用程序。S-BM基准作为代表基准测量的一种方式,可以用来评估多租用者应用程序和资源利用配置问题。通过两组设备对定制的负载生成器进行评测,包括在12个从小到大商业规模的代表负载中测量系统应用程序性能,以及在多样基础设施配置基础上对比其性能和不同负载成本等。实验研究负载与性能之间的关系,帮助找到用于多租用者应用程序的最优配置策略,结果表明,通过在共享环境下重新配置大和小交易应用程序,负载性能在同样资源成本情况下增长了30%。
To address the optimal configuration policy issues faced by multi-tenant applications,such as scale of operations and operational characteristics,we created a simulated vendor relationship management application called the S-BM benchmark which serves as a way of representing benchmark measurements that can be used to evaluate multi-tenant applications and resource utilization configuration issues. We measured the custom load generator across two sets of devices,including measuring the application performance of the system in 12 small to large commercial scale representative loads and comparing its performance against the cost of different loads based on a variety of infrastructure configurations. The results of these experiments which examined the relationship between load and performance helped us find the optimal configuration strategy for multi-tenant applications. In the end,we found that by reconfiguring large and small transactions applications in a shared environment,the performance of the load was increased by 30 percent at the same resource cost.
To address the optimal configuration policy issues faced by multi-tenant applications,such as scale of operations and operational characteristics,we created a simulated vendor relationship management application called the S-BM benchmark which serves as a way of representing benchmark measurements that can be used to evaluate multi-tenant applications and resource utilization configuration issues. We measured the custom load generator across two sets of devices,including measuring the application performance of the system in 12 small to large commercial scale representative loads and comparing its performance against the cost of different loads based on a variety of infrastructure configurations. The results of these experiments which examined the relationship between load and performance helped us find the optimal configuration strategy for multi-tenant applications. In the end,we found that by reconfiguring large and small transactions applications in a shared environment,the performance of the load was increased by 30 percent at the same resource cost.
引文
[1]杨建新.SAAS现状分析与前景展望[J].软件导刊,2012,11(1):1-2.
[2]郝晓宁.面向客户定制多租户SaaS应用的Benchmark系统的设计与实现[D].济南:山东大学,2014.
[3]许世网.SAAS的优缺点辨析[EB/OL].http://www.ccw.com.cn/cio/research/info/htm2007/20070314_245311.asp.
[4]GARCíA D F,GARCíA J.TPC-W E-commerce benchmark evaluation[C].Computer,2003:36-42.
[5]RUBIS.Rice university bidding system[EB/OL].http://rubis.ow2.org/.
[6]RUBBOS.Bulletin boardbenchmark[EB/OL].http://jmob.objectweb.org/rubbos.html.
[7]CECCHET E,MARGUERITE J,ZWAENEPOEL W.Performance and scalability of EJB applications[J].SIGPLAN,2002,37:246-261.
[8]BEITCH A,LIU B,YUNG T,et al.Rain:a workload generation toolkit for cloud computing applications[C].California:Technical Report UCB/EECS-2010-14,2010.
[9]DIWAKAR K.A synthetic workload generation technique for stress testing session-based systems[J].IEEE Transactions on Software Engineering,2006,32(11):868-882.
[10]HASHEM H N,ERIC R.Architectural contesting:exposing and exploiting temperamental behavior[J].SIGARCH Computer Architecture News,2007,35(3):28-35.
[11]DIWAKAR K.A Synthetic workload generation technique for stress testing session-based systems[J].IEEE Transactions on Software Engineering,2006,32(11):868-882.
[12]RAOUFEHSADAT H,DIWAKAR K,MARTIN F.Arlitt:web workload generation challenges-an empirical investigation[J].Software-Practice and Experience,2012,42(5):629-647.
[13]GUEYOUNG J,KAUSTUBH J,MATTI H,et al.A cost-sensitive adaptation engine for server consolidation of multi-tier applications[C].Illinois:ACM/IFIP/USENIX 10th International Middleware Conference(Middleware 2009),2009.
[14]WIKIPEDIA.Supplier relationship management[EB/OL].http://en.wikipedia.org/wiki/Supplier_relationship_management.
[15]SHARI L P.软件工程理论与实践(第2版)[M].吴丹,译.北京:清华大学出版社,2003.
[16]STEPHEN R.Scnach面向对象与传统软件工程[M].北京:机械工业出版社,2003.
[17]ZEYU D.S-BM:a benchmark suite for multi-tenant supplier relationship management service[C].Service Systems and Service Management(ICSSSM),2013:4-5.
[2]郝晓宁.面向客户定制多租户SaaS应用的Benchmark系统的设计与实现[D].济南:山东大学,2014.
[3]许世网.SAAS的优缺点辨析[EB/OL].http://www.ccw.com.cn/cio/research/info/htm2007/20070314_245311.asp.
[4]GARCíA D F,GARCíA J.TPC-W E-commerce benchmark evaluation[C].Computer,2003:36-42.
[5]RUBIS.Rice university bidding system[EB/OL].http://rubis.ow2.org/.
[6]RUBBOS.Bulletin boardbenchmark[EB/OL].http://jmob.objectweb.org/rubbos.html.
[7]CECCHET E,MARGUERITE J,ZWAENEPOEL W.Performance and scalability of EJB applications[J].SIGPLAN,2002,37:246-261.
[8]BEITCH A,LIU B,YUNG T,et al.Rain:a workload generation toolkit for cloud computing applications[C].California:Technical Report UCB/EECS-2010-14,2010.
[9]DIWAKAR K.A synthetic workload generation technique for stress testing session-based systems[J].IEEE Transactions on Software Engineering,2006,32(11):868-882.
[10]HASHEM H N,ERIC R.Architectural contesting:exposing and exploiting temperamental behavior[J].SIGARCH Computer Architecture News,2007,35(3):28-35.
[11]DIWAKAR K.A Synthetic workload generation technique for stress testing session-based systems[J].IEEE Transactions on Software Engineering,2006,32(11):868-882.
[12]RAOUFEHSADAT H,DIWAKAR K,MARTIN F.Arlitt:web workload generation challenges-an empirical investigation[J].Software-Practice and Experience,2012,42(5):629-647.
[13]GUEYOUNG J,KAUSTUBH J,MATTI H,et al.A cost-sensitive adaptation engine for server consolidation of multi-tier applications[C].Illinois:ACM/IFIP/USENIX 10th International Middleware Conference(Middleware 2009),2009.
[14]WIKIPEDIA.Supplier relationship management[EB/OL].http://en.wikipedia.org/wiki/Supplier_relationship_management.
[15]SHARI L P.软件工程理论与实践(第2版)[M].吴丹,译.北京:清华大学出版社,2003.
[16]STEPHEN R.Scnach面向对象与传统软件工程[M].北京:机械工业出版社,2003.
[17]ZEYU D.S-BM:a benchmark suite for multi-tenant supplier relationship management service[C].Service Systems and Service Management(ICSSSM),2013:4-5.