SPEC Seal of Reviewal SPECjEnterprise®2010 Result
Copyright © 2009-2010 Standard Performance Evaluation Corporation
SAP NetWeaver CE 7.1, MaxDB 7.8 on HP DL380 G6 X5570/X5560, Novell SLES10
1,323.33 SPECjEnterprise2010 EjOPS
Submitter:
HP
SPEC license # 3 Test date:
Mar-2010
Software Products
SAP Ehp1 for SAP NetWeaver CE 7.1 SP5
SAP JVM 5.1 64-bit Patch Level 49
SAP MaxDB 7.8.01.07 JDBC Driver
SAP MaxDB 7.8.01.10 64-Bit
Software Configurations
App Server SW Configuration
Emulator SW Configuration
Database SW Configuration
Driver SW Configuration
Hardware Systems
App Server HW
DB Server HW
Load Driver and Emulator HW

System Configuration Diagram

Benchmark
Modifications
Configuration
Bill of Materials
Other Info
General Notes
Full Disclosure Archive

SUT Configuration
JEE Server Nodes: 1   DB Server Nodes: 1
JEE Server CPUs: 8 cores, 2 chips   DB Server CPUs: 8 cores, 2 chips
JEE Instances: 4   DB Instances: 1
Other SUT Components:
   HP ProCurve Switch 3500yl-48G-PWR Intelligent Edge

Benchmark Settings
Dealer Injection Rate: 830   Ramp Up Time: 900 seconds
DB Load Injection Rate: 900   Steady State Time: 3,600 seconds
# of Dealer Agents: 8,300   Ramp Down Time: 300 seconds
# of Manufacturing Agents: 2,490  

Detailed Results
Dealer Transaction Mix Count Tx. Mix Failure Count
Purchase 752,006 25.02% 1
Manage 750,943 24.99% 1
Browse 1,502,473 49.99% 0
Total # Dealer Transactions 3,005,422
Manufacturing Transaction Mix Count Tx. Mix Failure Count
CreateVehicleEJB 877,642 49.91% 168
CreateVehicleWS 880,912 50.09% 169
Total # Manufacturing Transactions 1,758,554
Response Times Average Standard Deviation Maximum 90th% Reqd
Purchase 0.232 0.23 3.906 0.510 2.000
Manage 0.206 0.22 4.923 0.460 2.000
Browse 0.277 0.27 3.237 0.570 2.000
CreateVehicleEJB 0.144 0.11 2.193 0.280 5.000
CreateVehicleWS 0.165 0.12 2.615 0.310 5.000
Reproducibility Run (SPECjEnterprise2010 EjOPS) 1,323.64

SAP Ehp1 for SAP NetWeaver CE 7.1 SP5
Vendor: SAP Availability: Jun-2010
CTS Version: Java EE 5.0 Compatibility Test Suite patch B Date Passed CTS: Mar-2007
OS Name: Novell SLES10.2 (x86_64)
EJB Protocol: RMI/P4

SAP JVM 5.1 64-bit Patch Level 49
Vendor: SAP Availability: Feb-2010
OS Name: Novell SLES10.2 (x86_64)

SAP MaxDB 7.8.01.07 JDBC Driver
Vendor: SAP Availability: Oct-2009
OS Name: Novell SLES10.2 (x86_64)

SAP MaxDB 7.8.01.10 64-Bit
Vendor: SAP Availability: Jan-2010
OS Name: Novell SLES10.2 (x86_64)

App Server SW Configuration
JEE Product: SAP Ehp1 for SAP NetWeaver CE 7.1 SP5
JVM Product: SAP JVM 5.1 64-bit Patch Level 49
JDBC Product: SAP MaxDB 7.8.01.07 JDBC Driver
Hardware: App Server HW
# of Instances: 1
Interfaces:
Web Interface for the Order Domain
Web Services Interface for the Supplier Domain
EJB Interface for the Manufacturing Domain
Web Services Interface for the Manufacturing Domain
Notes / Tuning Information
Java engine system profiles as provided in the FDA (FDA/Config/AppServers/S01_J20_hpcc111, FDA/Config/AppServers/S01_J22_hpcc111).
Deployed and activated Zero Admin Template for SPECjEnterprise benchmark application as provided in the FDA (FDA/Deploy/template.sda).
   Among other things this Zero Admin Template automatically causes that the following JVM options are used:
      -Xmx5120m -Xms5120m -XX:NewSize=3584m -XX:MaxNewSize=3584m
      -verbose:gc -Xloggc:server.gc -Xss128k
      -XX:ReservedCodeCacheSize=200m -XX:CICompilerCount=10
      -XX:+UseParallelOldGC -XX:ParallelGCThreads=16
      -XX:SurvivorRatio=12 -XX:-UseAdaptiveSizePolicy
      -XX:-DoIoStatistics
Increased file descriptor limits of Linux user to 32768.
Applied SAP Notes:
   SAP Note 1298972 (Upgrade of System JDBC Driver and System Database of SAP Netweaver CE application server to MaxDB 7.8)
   SAP Note 1025085 (AdaptiveSizePolicy)
   SAP Note 1402539 (Usage of OldParallelGC)
Installation of SAP J2EE Kernel Part 1, Patch Level 76.
To use NUMA binding each Dialog Instance to a socket patch startsap of the SAP J2EE Kernel with startsap.diff of the FDA (FDA/Config/AppServers/startsap.diff).
Set the following parameters using the configuration editor mode of AS Java config tool:
   Parameters of JMS provider 'SPECj' (gets visible after specj.ear deployment):
      clientConsumerBuffer=10000000
      agentThreads=12
   Remove filter-mapping element from global-web.xml of servlet.jsp service.
   HTTP logging disabled by setting HttpHosts.Default.EnableLoging to false.
   Disable JPAMonitoring by setting OpenSQL.PropertysheetDatabaseStatistics.JPAMonitoring.JPAMonitoringActive to false.
   Disable OpenSQL Table Statistics by setting OpenSQL.PropertysheetTableStatistics.StatisticsActive to false.
   Disable OpenSQL Database Statistics by setting the following properties to false:
      OpenSQL.PropertysheetDatabaseStatistics.DefaultStatisticsOn
      OpenSQL.PropertysheetDatabaseStatistics.RequestStatisticsOn
      OpenSQL.PropertysheetDatabaseStatistics.TimeStatisticsOn

Emulator SW Configuration
JVM Product: SAP JVM 5.1 64-bit Patch Level 49
Web Product: SAP Ehp1 for SAP NetWeaver CE 7.1 SP5
Hardware: Load Driver and Emulator HW
# of Instances: 1
Notes / Tuning Information
Java engine system profile as provided in the FDA (FDA/Config/AppServers/EM1_J10_hpcc110).
Deployed and activated Zero Admin Template for SPECjEnterprise benchmark application as provided in the FDA (FDA/Deploy/template.sda).
   Among other things this Zero Admin Template automatically causes that the following JVM options are used:
      -verbose:gc -Xloggc:server.gc -Xss128k
      -XX:ReservedCodeCacheSize=200m -XX:CICompilerCount=10
      -XX:+UseParallelOldGC -XX:ParallelGCThreads=16
      -XX:SurvivorRatio=12 -XX:-UseAdaptiveSizePolicy
      -XX:-DoIoStatistics
After deploying the Zero Admin Template the heap size is reconfigured by setting the parameters
   initialHeapSize = maxHeapSize = 4096 and using default values for the new size of the heap.
Additional configuration as described in the "Note/Tuning" section of the "App Server SW Configuration".

Database SW Configuration
DB Product: SAP MaxDB 7.8.01.10 64-Bit
Hardware: DB Server HW
# of Instances: 1
Character Set: CESU-8
Notes / Tuning Information
Turn on log writing, turn off archiving.
Disable tracing by executing the dbmcli command 'trace_off all'.
Database parameters (different from default values):
   AutoLogBackupSize=70000
   CacheMemorySize=900000
   ClusterCompressionFillThreshold=0
   CommandBufferSize=49152
   DataCacheStripes=256
   EnableDataVolumeBalancing=NO
   EnableJoinHashTableOptimization=NO
   GarbageCollectors=10
   LoadBalancingCheckInterval=0
   LogQueueSize=4000
   MaxDataVolumes=40
   MaxLogVolumes=16
   MaxLogWriterTasks 8
   MaxUserTasks=500
   MCODIndicator=NO
   SessionTimeout=0
   TransactionLockManagementStripes=32
   TableLockManagementStripes=32
   UseMultiVersionReadSupport=yes (SAP Note 1384780)

Driver SW Configuration
JVM Product: SAP JVM 5.1 64-bit Patch Level 49
Hardware: DB Server HW
# of Instances: 1
Notes / Tuning Information
Please see run.xml provided in FDA for details (FDA/FinalRun/2416/run.xml). run.xml contains the following JVM options:
   -Xms1536M -Xmx1536M -Xss128k
   -verbose:gc -XX:+DisableExplicitGC
   -XX:+UseParallelOldGC -XX:ParallelGCThreads=16
   -XX:SurvivorRatio=12 -XX:-UseAdaptiveSizePolicy
   -Dsun.net.inetaddr.ttl=0 -Dnetworkaddress.cache.ttl=0 -Djava.net.preferIPv4Stack=true

App Server HW (SUT hardware)
Hardware Vendor: HP OS Vendor: Novell
Model Name: ProLiant DL380 G6 OS Name: Novell SLES10.2 (x86_64)
Processor: Intel(R) Xeon(R) CPU X5570 (Intel Turbo Boost Technology up to 3.33GHz) Filesystem: ext3
MHz: 2933 Disks: 2x146GB 15K RPM SFF SAS disks
# of CPUs: 8 cores, 2 chips, 4 cores/chip (8 HW threads/chip) Network Interface: 4x NetXtreme II BCM5709 Gigabit Ethernet
Memory (MB): 24576 Other Hardware: shared storage with DB Server HW (see "Other HW" of "DB Server HW")
DIMM[0] Count: 12 DIMM[0] Size: 2048
L1 Cache: 32kB shared instruction and data cache per core # of Systems: 1
L2 Cache: 256kB shared instruction and data cache per core H/W Available: Mar-2009
Other Cache: 8MB shared instruction and data cache per chip OS Available: May-2008
Notes / Tuning Information
BIOS Version P62 (10/01/2009) using the following options
  Power Management -> HP Power Profile -> Maximum Performance
  Power Management -> HP Power Regulator -> HP Static High Performance Mode
  Advanced Options -> Advanced Performance Tuning Options -> Memory Speed with 2 DIMMS per channel -> 1333 MHz Maximum

DB Server HW (SUT hardware)
Hardware Vendor: HP OS Vendor: Novell
Model Name: ProLiant DL380 G6 OS Name: Novell SLES10.2 (x86_64)
Processor: Intel(R) Xeon(R) CPU X5560 (Intel Turbo Boost Technology up to 3.33GHz) Filesystem: ext3
MHz: 2800 Disks: 2x146GB 15K RPM SSF SAS disks
# of CPUs: 8 cores, 2 chips, 4 cores/chip (8 HW threads/chip) Network Interface: 4x NetXtreme II BCM5709 Gigabit Ethernet
Memory (MB): 12288 Other Hardware: 1x StorageWorks 8/24 Base (16) Full Fabric Ports Enabled SAN Switch, 10x HP 8Gb Short Wave B-Series FC SFP+ 1 Pack, 1x HP StorageWorks 2024 Modular Smart Array 2.5-inch Drive Bay Chassis, 2x HP StorageWorks 2300fc G2 Modular Smart Array Controller, 24x HP 72GB 3G 15K 2.5-inch SAS SFF DP HDD
DIMM[0] Count: 6 DIMM[0] Size: 2048
L1 Cache: 32kB shared instruction and data cache per core # of Systems: 1
L2 Cache: 256kB shared instruction and data cache per core H/W Available: Mar-2009
Other Cache: 8MB shared instruction and data cache per chip OS Available: May-2008
Notes / Tuning Information
BIOS Version P62 (10/01/2009) using the following options
  Power Management -> HP Power Profile -> Maximum Performance
  Power Management -> HP Power Regulator -> HP Static High Performance Mode
  Advanced Options -> Advanced Performance Tuning Options -> Memory Speed with 2 DIMMS per channel -> 1333 MHz Maximum

Load Driver and Emulator HW (non-SUT hardware)
Hardware Vendor: HP OS Vendor: Novell
Model Name: ProLiant DL380 G6 OS Name: Novell SLES10.2 (x86_64)
Processor: Intel(R) Xeon(R) CPU X5560 (Intel Turbo Boost Technology up to 3.33GHz) Filesystem: ext3
MHz: 2800 Disks: 4x146GB 15K RPM SSF SAS disks
# of CPUs: 8 cores, 2 chips, 4 cores/chip (8 HW threads/chip) Network Interface: 4x NetXtreme II BCM5709 Gigabit Ethernet
Memory (MB): 12288 Other Hardware: shared storage with DB Server HW (see "Other HW" of "DB Server HW")
DIMM[0] Count: 6 DIMM[0] Size: 2048
L1 Cache: 32kB shared instruction and data cache per core # of Systems: 1
L2 Cache: 256kB shared instruction and data cache per core H/W Available: Mar-2009
Other Cache: 8MB shared instruction and data cache per chip OS Available: May-2008
Notes / Tuning Information
BIOS Version P62 (10/01/2009) using the following options
  Power Management -> HP Power Profile -> Maximum Performance
  Power Management -> HP Power Regulator -> HP Static High Performance Mode
  Advanced Options -> Advanced Performance Tuning Options -> Memory Speed with 2 DIMMS per channel -> 1333 MHz Maximum

Benchmark Modifications
Schema Modifications:
  Added Index on table S_PURCH_ORDER (see S_PURCH_ORDER.gdbtable in FDA/Deploy/DDIC.sda of the FDA)
  using the following columns - sorted ascending - in the given order:
     PO_SITE_ID, PO_START_DATE, PO_SUPP_ID
Load Program Modifications:
  no modifications

Benchmark Configuration Information
Isolation Requirement Info:
  The isolation level of the database connections is READ_COMMITTED -
  see data source descriptors data-sources.xml of FDA (FDA/Deploy/Descriptors/data-sources.xml).
  
  The SAPJPA implementation always performs JPA version checks against the database.
  This guarantees an isolation level of REAPEATABLE_READ where needed in the
  SPECjEnterprise benchmark application.
  
  The isolation level Strict REPEATABLE_READ and Strict READ_COMMITTED is achieved by
  disallowing Phantom Deletes on the database
  (database parameter UseMultiVersionReadSupport=yes).
  
  The database objects of the benchmark application are generated in the system database
  of the SAP Netweaver CE application server containing the tables for storing persistent
  JMS messages. The JMS provider uses the database connections of the application to store
  persistent JMS messages (see descriptor jms-resources.xml of FDA
  (FDA/Deploy/Descriptors/jms-resources.xml)) to meet the requirements of the Run Rules,
  chapter '2.10.2.1 JMS'.
  
  Please have a look at the FDA for the following configuration files which are bundled and
  deployed with the benchmark application archive specj.ear:
     data source configuration (FDA/Deploy/Descriptors/data-sources.xml)
     JPA descriptors (FDA/Deploy/Descriptors/persistence.xml, orm.xml)
     JMS configuration (FDA/Deploy/Descriptors/jms-resources.xml)
  E.g. the cache evict timeout of cached Item entities is 20 min by using the following
  configuration in the peristence.xml descriptor:
     com.sap.jpa.cache.query-result-cache.time-to-live=1200
Durability Requirement Info:
  To ensure database durability, RAID10 was used for the database logs maintained on the external storage.
  Two RAID10 disk arrays with each a total of 4x 73.4GB disks were used for the logs.
  Its cache has a battery backup capable of meeting the 24 hour run requirement.
Storage Requirement Info:
  After loading the database size was 16.2GB.
  The >60 minute run for this submission resulted in a DB size of 17.4GB, i.e. database size increased by max. 1.2GB per 1 hour.
  This extrapolates to less than 16.2GB + 1.2GB*24h = 45GB for a 24 hour period.
  Two RAID10 disk arrays with each a total of 8x 73.4GB disks were used to create the filesystem space for the database
   containing 28 data files with each a size 2200MB resulting in a database size of >60GB.

Bill of Materials
Qty   Supplier, Description                                                 Product #
----  -------------------------------------------                           ----------------
2x    HP DL380G6 X5560                                                      491315-421
2x    Intel Xeon Processor X5570 (2.93 GHz, 8MB L3 Cache, 95W,	            500094-B21
_     DDR3-1333, HT, Turbo 2/2/3/3)
6x    HP 2GB 2Rx8 PC3-10600R-9 Kit                                          500656-B21
4x    HP 146GB 3G SAS 15K SFF DP ENT HDD                                    504062-B21
2x    HP StorageWorks 82E 8Gb Dual-port PCI-e FC HBA                        AJ763A
+     HP Support - 3yr 24x7                                                 U4545E

1x    10622 G2 (22U) Rack Cabinet - Shock Pallet                            AF022A
2x    HP 7.3 kVA S132 Monitored PDU- single input, 10, 32A (INTL)           AF915A
1x    TFT7600 Rackmount Keyboard and Monitor INTL                           AG066A

1x    StorageWorks 8/24 Base (16) Full Fabric Ports Enabled SAN Switch      AM868A
10x   HP 8Gb Short Wave B-Series FC SFP+ 1 Pack                             AJ716A
1x    HP ProCurve Switch 3500yl-48G-PWR Intelligent Edge                    J8693A
+     HP Support - 3yr 24x7                                                 U2856E

1x    HP StorageWorks 2024 Modular Smart Array 2.5-inch Drive Bay Chassis   AJ949A
2x    HP StorageWorks 2300fc G2 Modular Smart Array Controller              AJ798A
24x   HP 72GB 3G 15K 2.5-inch SAS SFF DP HDD                                418371-B21
+     HP Support - 3yr 24x7                                                 UJ675E

1x    HP Novell SLES10.2 (x86_64) Media

5xCPU License SAP NetWeaver Foundation For Third Party Applications **      7009522
+     SAP MaxDB                                                             7002719
1     SAP Enterprise Support (3 years)                                      7003650

** multiplier for quadcore sockets is 2.5, i.e. 2.5 * 2 sockets = 5 CPUs


Other Benchmark Information
  DNS round robin load balancing was used and hosted on the database machine.
  All network connections were 1Gbps.

General Notes
The only errors in the driver log files were those that are
normally generated by this benchmark.

The data sources use 'SAP Open SQL for Java'. It guarantees
that all statements generated by the SAPJPA implementation
can be executed by all databases which are supported by SAP
and have the same semantical meaning.

According to the default value of 600 for the parameter
MaxSavepointTimeInterval the MaxDB database executes a
savepoint at least every 10 minutes. Hence at least two
database savepoints took place during the Measurement
Interval.


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

Copyright © 2009-2010 Standard Performance Evaluation Corporation

First published at SPEC.org on 25-Mar-2010