Setting LIBNAME Options That Affect Performance of SPD Engine Tables

specifies a list of paths in which to store partitions (.dpf files) for an SPD Engine table. The engine creates as many partitions as are needed to store all the data. The size of the partitions is set using the PARTSIZE= option. Partitions are created in the specified paths in a cyclic fashion. The data path area is best configured as multiple paths. Allot one I/O controller per data path to provide high I/O throughput, which is the rate at which requests for work are serviced by a computer system. The data path area is best configured for redundancy (RAID 1).

specifies a path or a list of paths in which to store the two index component files (.hbx and .idx) that are associated with an SPD Engine table. Additional specified paths accept the overflow from the immediately preceding path. The index path area is best configured as multiple paths. Use a volume manager file system that is striped across multiple disks (RAID 0) to enable adequate index performance, both when evaluating WHERE clauses and creating indexes in parallel. Redundancy (RAID 5 or RAID 10) is also recommended.

specifies a list of overflow paths in which to store metadata component (.mdf) files for an SPD Engine table. The metadata component file for each table must begin in the primary path. When that primary path is full, the overflow is sent to the specified METAPATH= location. The metadata path area is best configured for redundancy (RAID 1) so that metadata about the data and its indexes is not lost.

specifies the size (in megabytes) of the data component partitions when an SPD Engine table is created. By splitting the data portion of an SPD Engine table at fixed-size intervals, you can gain a high degree of scalability for some operations. For example, the SPD Engine can spawn threads in parallel, up to one thread per partition for WHERE evaluations.

specifies whether to create a temporary subdirectory of the directory specified in the Path field on the Library Properties wizard window. The directory is used to temporarily store the metadata component files associated with table creation. It is deleted at the end of the SAS session.

specifies that the SPD Engine should perform an automatic implicit sort when it finds a BY statement for processing data in the library (unless the data is indexed on the BY column). Valid values are YES (perform the sort) and NO (do not perform the sort). The default is YES.

specifies the number of the starting observation in a user-defined range of observations that are qualified with a WHERE expression. By default the SPD Engine processes all observations in the table.

specifies the number of the ending observation in a user-defined range of observations that are qualified with a WHERE expression. By default the SPD Engine processes all observations in the table.