建構在多雲端儲存供應商中之虛擬雲端儲存系統

Abstract

隨著雲端產業的快速發展，雲端儲存供應商提供使用者一個低成本且便利的網路空間，讓使用者可以有效地管理檔案或進行備份，但是在雲端儲存逐漸普及化之後，如果還是只依靠單一的雲端儲存供應商，越來越多的安全性問題將慢慢地浮出檯面，而我們要做的就是建立一個以磁碟陣列(RAID)為基礎的雲端儲存系統。 Papak系統是一個在不使用專門資料加密步驟的前提下仍能保有資料隱私的雲端儲存系統，我們模擬真實磁碟陣列的儲存結構與模式，並將資料分散地儲存在數個不同的雲端儲存供應商，使用此方法除了增加資料的安全性之外，更可以解決對單一雲端儲存供應商的依賴性和傳輸瓶頸。此外Papak系統更可以容忍雲端儲存供應商的暫時離線並持續保持資料的完整性，若原始本地主機不幸毀損，Papak系統亦可以支援異地重建的功能，讓使用者能在異地重新建立系統。 我們實作並測試了Papak系統，模擬了數種的使用情境並和其他演算法進行比較。實驗結果顯示，我們提出的演算法在處理大型檔案與多雲端空間整合兩方面有不錯的效果。

With the growth of cloud services, cloud storage is already part of our life. Cloud services provide user a cheap cloud storage to manage and backup data efficiently. As the cloud services become general, some security issues also follow if we still depend on single cloud storage. Thus we model an algorithm to simulate RAID (Redundant Array of Independent Disks) on cloud storage. Papak system is a cloud storage system which maintain data privacy without dedicated encryption but integrated with RAID simulation. We simulate a real RAID storage structure and pattern where data is split and uploaded to multiple cloud storages, this method increase data privacy, resolve dependence and transmission bottleneck of single cloud storage. Besides, Papak system also suffer cloud storage offline temporarily but keep data integrity. Papak system also support remote data rebuilding if necessary. We implement and evaluate our algorithm, simulate multiple situations and also compare with other algorithm. The results show our algorithm achieve better performance when process large files or numerous cloud storages are connected.