Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/51549
標題: 利用虛擬儲列達成主動式儲列管理之伺服器際公平服務
Achieving Inter-Server Fairness by Using Virtual Queues in AQM
作者: 林偉 
關鍵字: 資訊工程--硬體工程;基礎研究;active queue management;主動儲列管理;伺服器際公平性;虛擬儲列;封包排程;網路擁塞;inter-server fairness;virtual queues;packet scheduling;network congestion
摘要: 
A greedy or unresponsive server with multiple flows may contribute to unfair bandwidth sharing among servers, especially when the network becomes congested. We intend to examine the inter-server fairness problem in traditional AQM designs. We propose to design and develop new AQM algorithms for providing inter-server fairness services. The main idea behind our proposed research is to schedule packets from different servers into dedicated virtual queues. The dropping probabilities of the queues are determined independently by the occupancy of the individual queues. Our first step is to identify major traffic parameters of the inter-server fairness, and investigate their impact on the fairness of a congested network. Then we will take one further step to look into the inter-server fairness problem with an excessive number of flows. As the number of flows increases, the time and storage complexities of queue management grow exponentially. Under this circumstance, reducing the management complexities becomes a crucial design consideration in making a queue management algorithm practical.We aim to design efficient queue management algorithms specifically to keep the inter-server fairness in the case of excessive flows entering the network. To fully understand the effect of the traffic parameters, we plan to analytically develop closed-form equations for representing the inter-server fairness as a function of queue complexity. These closed-form equations are to serve as a guideline for the trade-off between desired inter-server fairness and queue complexity. With these equations, we will be able to predict accurately the inter-server fairness of the network for a given buffer configuration. Finally, using the ns-2 simulator, we will develop a set of simulations to verify the effectiveness of the queue management algorithms operating in various traffic conditions.

在擁塞的網路中,一個過份的伺服器可使用高流量數取得過多的頻寬,因而造成伺服器間不公平之頻寬分配。在此計劃我們將設計發展用以提供伺服器際公平服務之新演算法。我們將探討研究於傳統AQM中伺服器際公平性問題是如何產生。我們所提研究的主要想法是將來自不同伺服器的封包置於各自專有的虛擬儲列中,再將他們排程輸出。而這些儲列的丟棄率是分別依照儲列的占有率各自決定。此研究的第一步將是找出與伺服器際公平性的主要交通參數,並且研究這些參數對於一個壅塞網路的影響。然後我們的下一步是探討於大量流量數時的伺服器際公平性問題。當流量增加時,所需的儲列管理時間及空間複雜度也隨級數成長。在此狀況下,減少管理複雜度成為決定儲列管理是否實用的一個關鍵設計考量因素。我們研究主要目標是設計出有效率的儲列管理演算法,這些演算法將於超大流量進入網路時,用來維持伺服器際公平性。為能完全了解這些交通參數的影響效果,我們將以分析方法導出能表示伺服器際公平性之公式。這些公式可成為取捨所需伺服器際公平性與儲列複雜度間的決定原則。利用這些公式,我們將可精確地預測一個已知緩衝器組成之伺服器際公平性。最後,我們將利用ns-2模擬器發展出一組程式模擬,用以驗證這些儲列管理演算法於不同交通狀態時運作之效能。
URI: http://hdl.handle.net/11455/51549
其他識別: NSC97-2221-E005-054
Appears in Collections:資訊科學與工程學系所

Show full item record
 
TAIR Related Article

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.