Performance of encoding/decoding the transfer syntax

Abstract

Some basic performance results are reviewed that address the presentations layer aspect of open system interconnection (OSI) performance. In particular, an implementation of the encoding and decoding of the transfer syntax is evaluated based on testing performed in a US WEST Advanced Technologies laboratory. The tests indicate that response time comparisons for encoding/decoding single data types (character string, Boolean and integer) are nearly equal, but take roughly 30% longer to transfer than does the same sized block of octets without encoding/decoding. In general, the number of fields in the data set has a much greater impact on performance (with the ISODE implementation) than does the length of the data set. The ISODE software used to gather these performance measurements is primarily intended to proliferate use and experience with OSI protocols. Performance suffers as a result, especially for the implementation of the transfer syntax. It is highly likely that a production grade implementation of a transfer syntax based on the basic encoding rules (BER) would have substantially better performance. >