Tuesday Sep 27, 2011

SPARC T4-4 Server Sets World Record on PeopleSoft Payroll (N.A.) 9.1, Outperforms IBM Mainframe, HP Itanium

Oracle's SPARC T4-4 server achieved world record performance on the Unicode version of Oracle's PeopleSoft Enterprise Payroll (N.A) 9.1 extra-large volume model benchmark using Oracle Database 11g Release 2 running on Oracle Solaris 10.

  • The SPARC T4-4 server was able to process 1,460,544 payments/hour using PeopleSoft Payroll N.A 9.1.

  • The SPARC T4-4 server UNICODE result of 30.84 minutes on Payroll 9.1 is 2.8x faster than IBM z10 EC 2097 Payroll 9.0 (UNICODE version) result of 87.4 minutes. The IBM mainframe is rated at 6,512 MIPS.

  • The SPARC T4-4 server UNICODE result of 30.84 minutes on Payroll 9.1 is 3.1x faster than HP rx7640 Itanium2 non-UNICODE result of 96.17 minutes, on Payroll 9.0.

  • The average CPU utilization on the SPARC T4-4 server was only 30%, leaving significant room for business growth.

  • The SPARC T4-4 server processed payroll for 500,000 employees, 750,000 payments, in 30.84 minutes compared to the earlier world record result of 46.76 minutes on Oracle's SPARC Enterprise M5000 server.

  • The SPARC Enterprise M5000 server configured with eight 2.66 GHz SPARC64 VII processors has a result of 46.76 minutes on Payroll 9.1. That is 7% better than the result of 50.11 minutes on the SPARC Enterprise M5000 server configured with eight 2.53 GHz SPARC64 VII processors on Payroll 9.0. The difference in clock speed between the two processors is ~5%. That is close to the difference in the two results, thereby showing that the impact of the Payroll 9.1 benchmark on the overall result is about the same as that of Payroll 9.0.

Performance Landscape

PeopleSoft Payroll (N.A.) 9.1 – 500K Employees (7 Million SQL PayCalc, Unicode)

System OS/Database Payroll Processing
Result (minutes)
Run 1
Num of
SPARC T4-4, 4 x 3.0 GHz SPARC T4 Solaris/Oracle 11g 30.84 43.76 96
SPARC M5000, 8 x 2.66 GHz SPARC64 VII+ Solaris/Oracle 11g 46.76 66.28 32

PeopleSoft Payroll (N.A.) 9.0 – 500K Employees (3 Million SQL PayCalc, Non-Unicode)

System OS/Database Time in Minutes Num of
Run 1 Run 2 Run 3
Sun M5000, 8 x 2.53 GHz SPARC64 VII Solaris/Oracle 11g 50.11 73.88 534.20 1267.06 32
IBM z10 EC 2097, 9 x 4.4 GHz Gen1 Z/OS /DB2 58.96 80.5 250.68 462.6 8
IBM z10 EC 2097, 9 x 4.4 GHz Gen1 Z/OS /DB2 87.4 ** 107.6 - - 8
HP rx7640, 8 x 1.6 GHz Itanium2 HP-UX/Oracle 11g 96.17 133.63 712.72 1665.01 32

** This result was run with Unicode. The IBM z10 EC 2097 UNICODE result of 87.4 minutes is 48% slower than IBM z10 EC 2097 non-UNICODE result of 58.96 minutes, both on Payroll 9.0, each configured with nine 4.4GHz Gen1 processors.

Payroll 9.1 Compared to Payroll 9.0

Please note that Payroll 9.1 is Unicode based and Payroll 9.0 had non-Unicode and Unicode versions of the workload. There are 7 million executions of an SQL statement for the PayCalc batch process in Payroll 9.1 and 3 million executions of the same SQL statement for the PayCalc batch process in Payroll 9.0. This gets reflected in the elapsed time (27.33 min for 9.1 and 23.78 min for 9.0). The elapsed times of all other batch processes is lower (better) on 9.1.

Configuration Summary

Hardware Configuration:

SPARC T4-4 server
4 x 3.0 GHz SPARC T4 processors
256 GB memory
Sun Storage F5100 Flash Array
80 x 24 GB FMODs

Software Configuration:

Oracle Solaris 10 8/11
PeopleSoft HRMS and Campus Solutions 9.10.303
PeopleSoft Enterprise (PeopleTools) 8.51.035
Oracle Database 11g Release 2 (64-bit)
Micro Focus COBOLServer Express 5.1 (64-bit)

Benchmark Description

The PeopleSoft 9.1 Payroll (North America) benchmark is a performance benchmark established by PeopleSoft to demonstrate system performance for a range of processing volumes in a specific configuration. This information may be used to determine the software, hardware, and network configurations necessary to support processing volumes. This workload represents large batch runs typical of OLTP workloads during a mass update.

To measure five application business process run times for a database representing a large organization. The five processes are:

  • Paysheet Creation: Generates payroll data worksheets consisting of standard payroll information for each employee for a given pay cycle.

  • Payroll Calculation: Looks at paysheets and calculates checks for those employees.

  • Payroll Confirmation: Takes information generated by Payroll Calculation and updates the employees' balances with the calculated amounts.

  • Print Advice forms: The process takes the information generated by Payroll Calculations and Confirmation and produces an Advice for each employee to report Earnings, Taxes, Deduction, etc.

  • Create Direct Deposit File: The process takes information generated by the above processes and produces an electronic transmittal file that is used to transfer payroll funds directly into an employee's bank account.

Key Points and Best Practices

  • The SPARC T4-4 server with the Sun Storage F5100 Flash Array device had an average read throughput of up to 103 MB/sec and an average write throughput of up to 124 MB/sec while consuming 30% CPU on average.

  • The Sun Storage F5100 Flash Array device is a solid-state device that provides a read latency of only 0.5 msec. That is about 10 times faster than the normal disk latencies of 5 msec measured on this benchmark.

See Also

  • Oracle PeopleSoft Benchmark White Papers
  • PeopleSoft Enterprise Human Capital Management (Payroll)

  • PeopleSoft Enterprise Payroll 9.1 Using Oracle for Solaris (Unicode) on an Oracle's SPARC T4-4 – White Paper

  • SPARC T4-4 Server
  • Oracle Solaris
  • Oracle Database 11g Release 2 Enterprise Edition
  • Sun Storage F5100 Flash Array

Disclosure Statement

Oracle's PeopleSoft Payroll 9.1 benchmark, SPARC T4-4 30.84 min,
http://www.oracle.com/us/solutions/benchmark/apps-benchmark/peoplesoft-167486.html, results 9/26/2011.


BestPerf is the source of Oracle performance expertise. In this blog, Oracle's Strategic Applications Engineering group explores Oracle's performance results and shares best practices learned from working on Enterprise-wide Applications.

Index Pages

« April 2014