Reducing Transfer Delay with Dynamic Selection of Compression Formats

Chandra Krintz and Brad Calder

Abstract:

Internet computing is facilitated by the remote execution methodology in which programs transfer to a destination for execution. Since transfer time can substantially degrade performance of remotely executed (mobile) programs, file compression is used to reduce the amount that transfers. Compression techniques however, must trade off compression ratio for decompression time due to the algorithmic complexity of the former since the latter is performed at runtime in this environment.

With this work, we define Total Delay as the time for both transfer and decompression of a compressed file. To minimize total delay, a mobile program should be compressed in a format that minimizes total delay. Since both the transfer and decompression time are dependent upon the current, underlying resource performance, selection of the ``best'' format varies and no one compression format minimizes total delay for all resource performance characteristics. We present a system called Dynamic Compression Format Selection (DCFS) for automatic and dynamic selection of competitive, compression formats based on predicted values of future resource performance. Our results show that DCFS reduces 52% of total delay imposed by compressed transfer of Java archives (jar files) on average, for the networks, compression techniques, and benchmarks studied.