Troubleshooting BSAM Data Transfers

Data transfer rates using BSAM vary significantly from run to run even on the same system. For example, even with a high NCP, the transfer rate deteriorates when the I/O subsystem is moderately busy. If problems should occur, review these factors:

  • The REGION specified on the job card. With the higher number of I/O buffers comes the risk that if too many Processes run simultaneously, the above-the-line storage can be exhausted. To prevent out of storage abends, review both the MAXSTGIO initialization parameter and the job's region.
  • The number of concurrent transfers occurring on one DTF, even when all system components (CPU, DASD, CU, CHPID or network) run below capacity. For example, where a single transfer attains a transfer rate of 76 MB per second, two concurrent transfers potentially reduce it to 66 MB per second for each transfer, three transfers to 58 MB per second, and 4 transfers to 52 MB per second.
  • Hardware caching. When you run the same test case multiple times, usually the first runs slower than subsequent runs. For example, the initial transfer rate might be 38 MB/second, which increases to 75 MB/second for the second and subsequent runs.
  • Network. The transfer rate varies widely according both to the speed of the network and the volume of traffic on it.
  • Data set's device, CU, and CHPID configuration, speed, and how busy they are.
  • CPU speed, and how busy it is (sometimes a limiting factor).
  • Compression. Compression sometimes slows down the transfer due to extra CPU use.