No abstract

This paper presents an analytic model for investigating the throughput, delay and bu er utilization characteristics of partially ordered transport services. We analyze the e ects of packet and ack losses as well as applications' order requirements on overall system performance. The analytic model is veri ed by comparing its results against those of an OPNET simulation model. Analytic results show that for applications that can tolerate some reordering in the delivery of objects, use of partially ordered service instead of ordered service provides important bu er utilization and delay improvements, particularly as the loss rate increases and the order requirements of applications decrease. In terms of throughput, it makes no di erence which service i.e., ordered, partially ordered, unordered an application uses. Analytic study also shows that by judicious choice of sender's transmission order, overall system performance can further be improved in a partially ordered service.

No abstract

An analytic model is presented for a partially reliable transport protocol based o n r etransmissions. The model illustrates tradeo s between two QoS parameters delay and throughput, and various levels of reliability. The model predicts that the use of reliable transport service when an application only needs a partially reliable one causes considerable throughput decreases and delay increases in lossy networks. On the other hand, over lossy networks, unreliable transport service is unable to respect an application's loss tolerance. In lossy environments, partially reliable transport service avoids the extra c ost of reliable transport service, and, simultaneously, guarantees the minimal reliability that an application requires. Retransmission-based partially reliable transport service c an be p r ovided through either sender-based or receiverbased loss detection and recovery. Results show that both techniques provide almost identical reliability and delay. However, a sender-based approach provides better throughput than a r eceiver-based approach at high ack loss rates.

We describe an approach to transport QoS in unreliable networks that focuses on tradeoffs rather than guarantees. In particular, we investigate tradeoffs between qualitative QoS parameters such as order and reliability, and quantitative parameters such as delay and throughput.