SPEC Seal of Reviewal SPECjAppServer2001 Result
Copyright © 2002 Standard Performance Evaluation Corporation
IBM WebSphere Application Server on IBM eServer iSeries i890
804.08 BOPS@SingleNode
4,847.59
US$/BOPS@SingleNode
Submitter:
IBM Corporation
SPEC license # 11 Test date:
Sep-2002
EJB Container Avail: Jun-2002
EJB Container JVM Avail: Aug-2002
Software
EJB Container
Supplier Domain Container
Emulator Container
Database
JDBC
Other Software
Hardware
J2EE Application Server and Database Server
Load Driver and Emulator
Benchmark
Modifications
Configuration
Other Info
General Notes
Links
System Configuration Diagram

Full Disclosure Archive


Benchmark Settings
Orders Injection Rate: 480 Ramp Up Time: 600 seconds
# of Order Agents: 2 Ramp Down Time: 300 seconds
# of Manufacturing Agents: 3 Steady State Time: 1800 seconds
DB Load Injection Rate: 480 Trigger Time: 300 seconds
Total System Cost: US$3,897,874

Detailed Results
Order Transaction Mix Count Tx. Mix
New Order 432,110 50.02%
Change Order 172,668 19.99%
Order Status 172,782 20.00%
Customer Status 86,395 10.00%
Total # Order Transactions 863,955
Total # Manufacturing Transactions 583,398
Response Times Average Maximum 90th% Reqd
New Order 0.27 16.45 0.70 2.00
Change Order 0.14 16.41 0.30 2.00
Order Status 0.07 15.47 0.10 2.00
Customer Status 0.11 15.47 0.20 2.00
Manufacturing 2.48 94.99 4.50 5.00
Reproducibility Run (BOPS) 805.15

EJB Container
Name: WebSphere Application Server 4.0.3, Advanced Edition JVM Name: iSeries Developer Kit for Java 1.3.1 (64-bit)
Vendor: IBM Corporation JVM Vendor: IBM Corporation
Available: Jun-2002 Available: Aug-2002
Instances: 1
Passed CTS: Oct-2001
Protocol: RMI/IIOP
Tuning Information
Application Server module visibility: Application
EJB Container Service cache size: 74707
EJB Container Service cache clean-up interval: 3000 (ms)
Object Request Broker thread pool size: 150
JVM Tuning Information
System Properties:
    javax.rmi.CORBA.UtilClass=com.ibm.CORBA.iiop.Util
    com.ibm.ws.OrbThreadPoolGrowable=false
    com.ibm.CORBA.LocalHost=10.1.1.11 (to use GB ethernet line)
    java.compiler=jitc
    os400.jit.mmi.threshold=50
Initial Java heap size = 2048 MB
Maximum Java heap size = *NOMAX (default value)

Supplier Domain Container
Name: WebSphere Application Server 4.0.3, Advanced Edition JVM Name: iSeries Developer Kit for Java 1.3.1 (64-bit)
Vendor: IBM Corporation JVM Vendor: IBM Corporation
Available: Jun-2002 Available: Aug-2002
Tuning Information
Web Container Service settings:
    minimum thread pool size: 150
    maximum thread pool size: 150
    HTTP transports: *:9080
    maximum keep alives: 0
Virtual Host settings:
    name: default_host
    aliases: *:9080
JVM Tuning Information
System Properties:
    java.compiler=jitc
    os400.jit.mmi.threshold=50
Initial Java heap size = 2048 MB
Maximum Java heap size = *NOMAX (default value)

Emulator Container
Name: WebSphere Application Server 4.0.3, Advanced Edition JVM Name: iSeries Developer Kit for Java 1.3.1 (64-bit)
Vendor: IBM Corporation JVM Vendor: IBM Corporation
Available: Jun-2002 Available: Aug-2002
Tuning Information
Web Container Service settings:
    minimum thread pool size: 150
    maximum thread pool size: 150
    HTTP transports: *:9080
    maximum keep alives: 0
JVM Tuning Information
System Properties:
    java.compiler=jitc
    os400.jit.mmi.threshold=50
Initial Java heap size = 512 MB
Maximum Java heap size = *NOMAX (default value)

Database
Name: DB2 UDB for IBM iSeries (V5R2)
Vendor: IBM Corporation
Available: Aug-2002
Tuning Information
All drives were configured as a single logical disk array, with RAID-5 protection.
     The Operating System, WebSphere, and the SPECjAppServer2001 database (including
     the database journal and all journal receivers) resided on this array of
     disks.

JDBC
Name: iSeries Developer Kit for Java Native JDBC Driver (V5R2)
Vendor: IBM Corporation
Available: Aug-2002
Tuning Information
JDBC implementation class: com.ibm.db2.jdbc.app.DB2StdConnectionPoolDataSource
    This is the iSeries "Native JDBC Driver", and is a Type 2 JDBC driver.
databaseName: *LOCAL/spec
SPECDB data source
    Minimum pool size: 150
    Maximum pool size: 150
    Statement cache size: 12000
SPECUtilDB data source (used for the SequenceEnt bean)
    Minimum pool size: 30
    Maximum pool size: 30
    Statement cache size: 150

Other Software
Name:
Vendor:
Available:
Tuning Information


J2EE Application Server and Database Server (1 system)
Hardware Vendor: IBM Corporation OS Vendor: IBM Corporation
Model Name: 890-0198 OS Name: OS/400 Version 5 Release 2
Processor: Power4 Filesystem: IFS
MHz: 1300 Disks: 45x17.54 GB SCSI
# of CPUs: 32 cores, 32 chips, 1 core/chip Network Interface: 1000BASE-SX (Gigabit) Ethernet
Memory (MB): 65536 Other Hardware:
L1 Cache: 64KBI+32KBD (on chip) per CPU # of Systems: 1
L2 Cache: 1440KB unified (on chip), shared per two CPUs H/W Available: Aug-2002
Other Cache: 128MB unified (off chip) per MCM (8 processors per MCM) OS Available: Aug-2002
Notes / Tuning Information
The following user-configurable system values were changed from their shipped settings:

                Current    Shipped
    Name        value      value     Description
    ----------  ---------  --------  ---------------------------------------
    QBASACTLVL  32267      6         Base storage pool activity level
    QDBFSTCCOL  *NONE      *ALL      Database file statistics collection
    QMCHPOOL    4192064    20000     Machine storage pool size
    QPFRADJ     0          2         Performance adjustment
    QQRYDEGREE  *OPTIMIZE  *NONE     Parallel processing degree
    QSECURITY   20         40        System security level
    EDTRCYAP    *NONE      70        System access path recovery time

The following Group PTFs were applied:
    OS/400: C2211520 + SI05952
    Java:   SF99169-01 + MF29435

The measured and priced configurations are equivalent.

Load Driver and Emulator (1 system)
Hardware Vendor: IBM Corporation OS Vendor: IBM Corporation
Model Name: 830-2403 OS Name: OS/400 Version 5 Release 2
Processor: PowerPC Istar Filesystem: IFS
MHz: 540 Disks: 60x17.54 GB SCSI
# of CPUs: 8 cores, 8 chips, 1 core/chip Network Interface: 1000BASE-SX (Gigabit) Ethernet
Memory (MB): 32 GB Other Hardware:
L1 Cache: 128KB(I)+128KB(D) # of Systems: 1
L2 Cache: 4MB(I+D) off chip H/W Available: May-2000
Other Cache: OS Available: Aug-2002
Notes / Tuning Information
The following user-configurable system values were changed from their shipped settings:

                Current    Shipped
    Name        value      value     Description
    ----------  ---------  --------  ---------------------------------------
    QBASACTLVL  32267      6         Base storage pool activity level
    QMCHPOOL    3300012    20000     Machine storage pool size
    QPFRADJ     0          2         Performance adjustment
    QSECURITY   20         40        System security level

The following Group PTFs were applied:
    OS/400: C2211520
    Java:   SF99169-01

Driver JVM used: iSeries Developer Kit for Java 1.3.1 (64-bit)

Benchmark Modifications
Schema Modifications:
  The following indexes where added to the default schema to improve the
  performance of queries executed by the benchmark:
      create index SPEC/M_LO_IDX on SPEC/M_LARGEORDER (LO_ID)
      create index SPEC/M_LO_IDX2 on SPEC/M_LARGEORDER (LO_O_ID, LO_OL_ID)
      create index SPEC/O_ORDL_IX2 on SPEC/O_ORDERLINE (OL_O_ID)
      create index SPEC/O_ORDL_IX3 on SPEC/O_ORDERLINE (OL_O_ID, OL_I_ID)
Load Program Modifications:
  The load programs were not modified.
Reference Bean Modifications:
  No changes were made to the reference beans.

Benchmark Configuration Information
Persistence Mode Used:
  All beans were deployed using Container Managed Persistence (CMP).
Isolation Requirement Info:
  All beans were deployed with the isolation level set to REPEATABLE_READ. DB2
  translates this isolation level to Read Stability.  Read Stability ensures
  that all rows referenced in a transaction have a read-lock associated with
  them to ensure that other transactions cannot change the rows until the first
  referencing transaction completes.
  
  The following bean finder methods were deployed to use a FOR UPDATE clause
  on the SELECT during bean hydration:
  
      Mfg:
          Inventory.findByPrimaryKey
      Orders:
          Order.findByCustomer
          Order.findByPrimaryKey
          Orderline.findByOrder
          Orderline.findByOrderAndItem
          Orderline.findByPrimaryKey
      Supplier:
          SComponent.findByPrimaryKey
          PO.findByPO
          POLine.findByPO
          POLine.findByPrimaryKey
      Util:
          Sequence.findByPrimaryKey
Durability Requirement Info:
  All drives on the system were configured with RAID-5.  Database changes were
  journalled to disk prior to the completion of each transaction commit.
Storage Requirement Info:
  A 45 minute run at an injection rate of 480 increased storage use by 6.8 GB.
  Given linear scaling, eight hours of run time would at that rate will require
  72.6 GB of space.  The system was configured with 789.3 GB of disk storage
  (631.4 GB after enabling RAID-5 protection).  Over 560 GB of free space was
  available prior to each run.
Argument Passing Semantics:
  WebSphere Application Server uses pass-by-value semantics by default.  This can
  be confirmed in WebSphere's output when starting the application server by the
  following message:
       JORB0012: Pass by reference has been set to:  false (NoLocalCopies = false)

Other Benchmark Information
  No software or hardware was used to influence the flow of network traffic beyond
      basic IP routing and switching. No load balancing was performed.
  The driver and SUT were directly connected by a single GB ethernet line.
  The xerces.jar package in the jars sub-directory of the SPECjAppServer2001 kit
      was not used. IBM WebSphere Application Server 4.0.3 is packaged and
      tested with xerces.jar version 1.4.2, so it was used in all runs.

General Notes
No errors were reported during the final or reproducability runs.


For questions about this result, please contact the submitter: IBM Corporation
For other inquiries, please contact webmaster@spec.org

Benchmark run on Sun Sep 01 13:53:43 CDT 2002 by SPECjAppServer2001 v1.12
Result submitted on Thu Sep 05 03:03:00 EDT 2002
Report generated by SPECjAppServer2001 Reporter v1.09

Copyright © 2002 Standard Performance Evaluation Corporation

First published at SPEC.org on 18-Sep-2002