一個往返時間驅動的雲端資源管理框架

隨著軟體即服務的軟體交付方式愈來愈流行，供應商不僅需要開發出符合需求的功能，也要負責管理其軟體的運作狀況，其服務品質必須符合提供的服務水準協議(Service Level Agreement)。而利用雲端運算服務的彈性，使系統資源隨著當下的使用流量進行縮放變成可能,讓服務供應商能夠更好的保證其服務達到目標水準，同時減少運算資源的浪費。一般關於資源自動增減的系統與研究，皆嘗試利用各種於服務端取得的系統狀態來預測、估算使用者端可能感受到的服務品質，並作為增減的依據。但因現在的雲端運算服務皆以虛擬化技術提供資源，這意味著我們無法知道資源確切的能力為何，會受到共享同一實體機器的其他資源和Hypervisor的策略影響產生差異，且服務端與使用者端間的網路品質是會變化的，這使得要在各種情況下準確估算使用者端的服務品質變得困難。本研究提出一創新的自動運算資源管理系統，直接從終端量測使用者觀點的任務往返時間，並作為調節運算資源規模的依據。另外，也以此機制為基礎，提供自動化建立自動增長計畫的工具，應用於具有規律流量變化的服務，能夠依據過去服務流量紀錄預先進行測試，提前進行資源預備。本研究以一個檔案上傳服務對系統進行驗證，從實驗結果看,此系統確實能以往返時間有效的管理資源。而若進行預先測試規劃，則又能進一步提高服務表現。

關鍵字

自動擴增；水平增長；回應時間；雲端運算

並列摘要

A provider of Software as a service must not only focus on software functionality, but also to assure the service management, data security and execution performance, etc. The service quality needs to fulfill the Service Level Agreement (SLA). Today, service provider can use Infrastructure as a Service (IaaS) to obtain resource needed for serving their customer. By pay-as-you-go manner, service provider can save a lot cost for maintain the idle resource. Auto scaling is a general mechanism to using the IaaS. Previous products or studies monitor the system status, such as CPU Utilization, disk I/O, and network I/O, on server-side to make decision of when to adjust the amount of resource. But it can’t know the performance of real user feeling. To ensure the stability of service performance, in this thesis, we proposed two novel auto-scaling mechanisms. For first mechanism, we setup some response time monitor on client-side. It uses the information obtained from client side to drive a dynamic auto scaling operation. For second mechanism, we implement a schedule-based auto scaling pre-configuration maker to produce a pre-testing for a workload history and to find what amount of resource needed on each time. The result of experiment shows that the system is effective to reach the objective. And it can handle some quality variation that only monitoring server-side information can’t be figured out.

並列關鍵字

auto scale ； scale out ； response time ； cloud computing

參考文獻

[1] M. Cusumano, “Cloud Computing and SaaS As New Computing Platforms,” Commun ACM, vol. 53, no. 4, pp. 27–29, Apr. 2010.

[4] T. Lorido-Botrán, J. Miguel-Alonso, and J. A. Lozano, “Auto-scaling Techniques for Elastic Applications in Cloud Environments,” Department of Computer Architecture and Technology, UPV/EHU, EHU-KAT-IK, 2012.

[6] “RightScale: Cloud Portfolio Management by RightScale,” RightScale: Cloud Portfolio Management by RightScale. [Online]. Available: http://www.rightscale.com.

[18] M. Mao and M. Humphrey, “Scaling and Scheduling to Maximize Application Performance within Budget Constraints in Cloud Workflows,” in 2013 IEEE 27th International Symposium on Parallel Distributed Processing (IPDPS), 2013, pp. 67–78.

[19] H. Kang, J. Koh, Y. Kim, and J. Hahm, “A SLA driven VM auto-scaling method in hybrid cloud environment,” in Network Operations and Management Symposium (APNOMS), 2013 15th Asia-Pacific, 2013, pp. 1–6.

國際替代計量

一個往返時間驅動的雲端資源管理框架

全文下載

主題瀏覽