Based on the characteristics of TSS batch processing, Trillium Software recommends the following configuration to obtain optimum performance:
-
Configure the disk and controllers to perform simple flat-file reads/writes for the maximum throughput. The specific technique varies by the environment. For example, if TSS is running on a physical server, we recommend the drive(s) that handle the batch I/O be attached directly to that machine and not through the network. If the servers where TSS is running in Production are virtual machines, configure the disk allocation/file system and controllers to read and write flat files as quickly as possible with a minimum of network traffic.
-
Use pipes to allow a TSS script to handle more records, more quickly by reducing the number of disk writes between processes. See Using Pipes.
-
Consider using PVSCSI (Paravirtual) controllers, one for the data and a separate one for any sortwork files.
-
Create separate mounts by project and split the data I/O from the sortwork space within each project.
-
Configure TSS to read the large postal directory files into memory upon startup.