products ONsalessupportpartnersnewscompany  
Search
productsdmxBackup Expressss ON
SyncSort for Unix, LINUX & Windows
SyncSort for z/OS
Features and Benefits
Technical Specifications
Visual SyncSort for z/OS
MVS to UNIX Sort Converter
PipeSort
Features and Benefits
Technical Specifications
PROC SYNCSORT
SyncSort VSE
FilePort
News, Events, Training
Case Studies, White Papers, Articles
Customer Testimonials
Contact Us

PipeSort Technical Specifications

PipeSort is an innovative software product that works with SyncSort to run multiple sorts simultaneously. PipeSort reads the input data only once to produce differently sorted output data sets.

PipeSort can cut total elapsed time by more than 50 percent compared to running consecutive sorts.

PipeSort performs up to eight simultaneous sorts and outputs a file for each set of sort keys (SORT control statements). The complete range of SyncSort control statements and PARMs is available for the individual sort operations. Thus, in addition to different sequencing, the records in each output file can be uniquely selected and reformatted (with INCLUDE/OMIT and INREC/OUTREC control statements.)

Elapsed time savings

PipeSort exploits the rich multi-programming potential of MVS, OS/390, and z/OS to cut total elapsed time by replacing serial with parallel sort execution.

Parallel Sort Execution Cuts Elapsed Time

Benchmark tests on a one-gigabyte DASD file sorted in three different sequences compared consecutive executions versus PipeSort. The consecutive executions required 1 hr 9 minutes total elapsed time. PipeSort needed only 31 minutes, a savings of 55%. Even greater savings can be achieved when PipeSort replaces more individual sorts.

PipeSort reads the input file only once and distributes the input records to multiple simultaneous SyncSort executions. This method enables PipeSort to eliminate the considerable I/O overhead involved in reading the same input data for every sort.

Resource requirements

You must have adequate resources to run PipeSort successfully. PipeSort requires all the resources (DASD space and central storage) that would have been spread over the individual sort executions. For example, if PipeSort combines five sorts that each require 50 cylinders of DASD, PipeSort will need 250 cylinders. Also, CPU time may increase compared to the sum of CPU times for individual sorts.

Compatibility and implementation

PipeSort requires SyncSort and is easily installed. SyncSort control statements in existing applications can be used without change.

Implementation is through simple JCL that includes DD statements specific to PipeSort; for example, SRTnCNTL DD statements replace SYSIN DD statements. The following example shows PipeSort JCL for two sorts:

PipeSort JCL for Two Sorts

PipeSort may also be invoked from a user program, which may have input/output exits.

Product Literature

Download the PipeSort Technical Specifications

Free Trial

To request a FREE, FULL-FUNCTIONING EVALUATION COPY, or for more information about PipeSort, please contact us.
CopyrightPrivacy StatementPrivacy StatementPrivacy Statement