共用緩衝記憶體交換元之進階服務設計

Abstract

反應式頻寬仲裁法為一種有效仲裁各個優先權等級或單點和多點傳輸間服 務競爭之方法; 區域性推出式細胞漏失控制法則為一種可被實現及有效率 之緩衝區管理方法.反應式頻寬仲裁法結合了頻寬配置法及反應式仲裁法, 頻寬配置比是根據目前之傳輸特性及服務品質之需求來分配, 而細胞分送 之方法則考慮了頻寬比及佇列狀態. 與頻寬配置法比較, 反應式頻寬仲裁 法可動態地反應目前傳輸之變動情形; 與反應式仲裁法比較, 反應式頻寬 仲裁法即使在暫時性不平衡之傳輸狀態亦能公平地服務各個優先權等級. 使用反應式頻寬仲裁法, 我們可有系統地利用解線性方程式之方法來設定 系統參數以達成各種服務品質之需求. 此外單點和多點傳輸間之服務仲裁 亦可以反應式頻寬仲裁法來達成, 與優先權等級相似, 可以系統地方法來 符合各種服務品質之需求. 由於服務品質尚包括細胞漏失率, 所以在交換 元中緩衝區之管理方法是必備的, 區域性推出式細胞漏失控制法可避免拋 棄佇列中最後一個細胞而導致系統停頓之現象. 根據模擬, 區域性推出式 細胞漏失控制法在可實現之小型搜索區間即可有效地改善細胞漏失率. 反 應式頻寬仲裁法及區域性推出式細胞漏失控制法均併入通用型佇列管理之 硬體設計中, 對四個優先權等級, 傳輸率 622 MHz 之 8x8交換元其佇列 管理器之複雜度為 130K 個邏輯閘. A new service scheduling scheme, reactive bandwidth arbitration (RBA),is proposed as an effective way to arbitrate the contention among priority classes or between unicast and multicast services in ATM switches. And a new loss priority policy, local pushout discarding (LPD), is proposed as an effective and implementable way on buffer management.The RBA scheme integrates two previously proposed concepts: bandwidth allocation and reactive arbitration. Bandwidth ratios are allocated at call setup based on the traffic characteristics and service requirements. The arbitration of cell delivery takes into account the allocated bandwidth ratio and the queue status. Comparing to bandwidth allocation, RBA is capable of scheduling the service dynamically to react to the traffic fluctuation. Comparing to reactive arbitration, RBA can more fairly serve different service types even undertemporarily imbalanced traffic conditions. We found that by bandwidth allocation, a set of queue threshold levels can be obtained to provide a good approximation to the desired delay performance. Furthermore, around this initial configuration, a linear relationship between the queue threshold and the resulting delay performance can be established to fine-tune the configuration parameters for the desired delay performance.When arbitrating the contention between unicast queues andthe shared multicast queue, RBA is also effective to achieved the desired delay performance. Furthermore, RBA is capable ofdelivering more than one multicast cell in every cell slot.In order to meet the QoS requirements on cell loss rate, a new buffermanagement policy LPD is introduced. The LPD scheme searches for the local longest output queue for cell discarding. It avoids discarding the last cell of the output port and the subsequent idle cycle. By simulationresults, the LPD scheme is capable of achieving significantimprovement on cell loss performance with small local searching range. The proposed RBA and LPD schemes are incorporated in our generalpurpose queue manager architecture for shared buffer ATM switches. For an 8x8 switch with four priority classes per port and link rate at 622 Mbps,the queue manager complexity is about 130k gates, and a single chip implementation is currently being attempted.