基於OpenStack雲端系統結合Ceph軟體定義分散式儲存系統效能優化之研究

近年來資料量暴增、雲端運算技術普及與物聯網應用帶動下，資料儲存的處理需求足年倍增，導致許多組織與企業紛紛建置資料中心。由於傳統的伺服器架構限制，資源通常難以有效地利用，而導致研究的效能損失、缺乏資料安全性與可靠性。有鑑於此，基於軟體定義式基礎架構的發展被紛紛提出，其中以開源碼OpenStack雲端系統最受企業關注，其提供了基於軟體定義式的運算、網路與儲存等模型，使資源經過虛擬化方便以軟體進行彈性分配與使用。然而，OpenStack中虛擬化儲存服務預設下缺乏統一的儲存後台，造成儲存系統缺乏管理性與彈性，同時自動容錯與擴展性也受到挑戰，因此，許多企業採用了開源碼Ceph作為OpenStack儲存統一後台。本研究提出一基於OpenStack雲端系統結合Ceph軟體定義分散式儲存系統效能優化方法，針對Ceph儲存系統進行混合儲存放置規則設計與分層快取儲存優化建置，並搭配Linux核心快取技術優化儲存系統，提供一高效能儲存服務解決方案。本研究基於論文提出之三種優化方法進行組合觀察效能差異，並且透過不同混合儲存放置規則等，可能影響儲存效能之相關因素進行實驗，藉此評估與分析最佳的參數設定與優化方法組合，實驗結果顯示建立與設計一混合儲存放置規則，將對儲存效能進行最佳的分配與利用。

關鍵字

OpenStack ； Ceph ；軟體定義儲存；雲端運算；分層快取

並列摘要

In recent years, due to explosive increase in data volume, popularization of cloud computing and the need of internet of things, the processing demand of data storage was in double increase year by year, which led to higher and higher complication in the data centers built by many organizations and enterprises. Due to the limitation of traditional server architecture, the resources were usually difficult to be effectively used, which led to serious effectiveness loss and the lack of data security and reliability. Therefore, to deal with such issue, development based on software-defined basic architecture was proposed one after another, among them, open source OpenStack cloud system caught the greatest attention from the enterprise. However, in OpenStack, the defaulted virtualized storage service was lack of universal storage platform, which led to the lack of managerial characteristic and flexibility in the storage system, meanwhile, indexes such as automatic fault tolerance and ease of expandability were also challenged, therefore, to cope with such issue, open source Ceph is adopted as the storage system of OpenStack. In this study, methods for optimization of the performance of OpenStack cloud system in association with Ceph software-defined distributed storage system were proposed to make storage placement rule design and layered cache storage optimization setup on Ceph storage system. This study has made further investigation and evaluation on Ceph distributed storage system, three optimization methods were proposed in this study to observe the effectiveness differences through different combinations, meanwhile, experiments were conducted on related factors that might possibly affect the storage effectiveness such as different mixed storage placement rule so as to evaluate and analyze optimized parameter setup and optimized method combination. Experimental result showed the proposed storage optimization techniques will make the system to achieve optimized storage effectively.

並列關鍵字

OpenStack ； Ceph ； Software-defined Storage ； Cache Tiering ； Cloud Computing

參考文獻

[4] 鐘嘉德、高天助、楊嘉栩(2010)：雲端運算與產業發展。研考雙月刊，34(4)，20-31.

[13] 蔡權昱, and 蔡錫鈞. NCTU CStack: OpenStack 與 Ceph 的整合與應用. Diss. 2013.

[44] Konstantin Shvachko , Hairong Kuang , Sanjay Radia , Robert Chansler, "The Hadoop Distributed File System", Proceedings of the 2010 IEEE 26th Symposium on Mass Storage Systems and Technologies (MSST), p.1-10, May 03-07, 2010.

[82] Srinivasan, Mohan, Paul Saab, and V. Tkachenko. "Flashcache." (2010).

[1] Mell, Peter, and Tim Grance. "The NIST definition of cloud computing." (2011).

國際替代計量

基於OpenStack雲端系統結合Ceph軟體定義分散式儲存系統效能優化之研究

全文下載

主題瀏覽