一合作運行之分散式記憶體快取系統

Abstract

記憶體快取是設計將資料和物件暫存到記憶體以減少額外的資料讀取次數如資料庫和磁碟 I/O以增進系統效能。Memcached，是一個由Danga Interactive設計給LiveJournal且開放原始碼的記憶體快取的實作，提供了有效率的單台機器的記憶體快取功能。然而，一台 32位元的機器頂多有4GB的記憶體，這些記憶體對大規模的系統來說是不足的。在這篇論文裡，我們根據memcached設計一個有彈性的分散式記憶體快取系統的架構以解決記憶體容量不足的問題，並且為了管理伺服器及平衡工作量設計了一個分散式的網路給記憶體快取伺服器們。實驗的結果展示了當使用相同的分散方法時，我們能給予相較於傳統方法較佳的系統效能。此外，較少的記憶體被逐出的次數證明了我們能有效的利用伺服器們的記憶體

Memory Cache is designed to speed up the system performance by caching data and objects in memory to reduce the number of times an external data source must be read such as database and the amount of disc I/O. Memcached, an open-source implementation of memory cache developed by Danga Interactive for LiveJournal, provides efficient memory cache function on a single machine. However, there is at most 4GB memory in a 32-bit machine and it does not meet the requirement of the large-scale systems. In this paper, we design a scalable distributed memory cache system architecture based on memcached to solve the problem of limited memory capacity and designs a decentralized network of memory cache server to manage the servers and balance the workload. Experimental results show that we provide a better system performance than using traditional method when a client uses the same distribution method. Besides, the fewer evictions proves that we can effectively use the memory of the servers.