Design, implementation, and performance measurement of a native-mode ATM transport layer (extended version)

We describe the design, implementation, and performance measurement of a transport layer targeted specifically for asynchronous transfer mode (ATM) networks. The layer has been built from scratch to minimize overhead in the critical path, provide per-virtual circuit quality of service (QoS) guarantees, and take advantage of ATM adaptation layer 5 functionality. It provides reliable and unreliable data delivery with a choice of feedback and leaky-bucket flow control. These services can be combined to create per-virtual-circuit customized transport services. Our work is novel in that it provides high-performance, reliable, flow-controlled transport service using cheap personal computers (PCs). We describe the mechanisms and the operating system support needed to provide these services in detail. An extensive performance measurement allows us to pinpoint and eliminate inefficiencies in our implementation. With this tuning, we are able to achieve a user-to-user throughput of 55 Mb/s between two 66 MHz Intel 80486 personal computers with FORE Systems´ HPA-200 EISA-bus host adaptors. The user-to-user latency for small messages is around 720 μs. These figures compare favorably with the performance of far more expensive workstations and validate the correctness of our design choices