BM Seer Unofficial thoughts from an anonymous Sun employee

SPECjAppServer2004 World Record Single Application Server: Sun SPARC Enterprise T5240

Wednesday Apr 09, 2008

The Sun SPARC Enterprise T5240 server delivered a World Record single server result of 3331.31 SPECjAppServer2004 JOPS@Standard. This result used one Sun SPARC Enterprise T5240 (two UltraSPARC T2 Plus 1.4GHz chips) for the application server and a Sun SPARC Enterprise T5240 (with two UltraSPARC T2 Plus 1.4GHz chips) for the database server.

This benchmark used the Oracle Application Server 10g and Oracle Database 10g Enterprise Edition. This benchmark result proves that the Sun SPARC Enterprise T5240 server using the UltraSPARC T2 Plus processor performs as an oustanding J2EE application server as well as an Oracle 10G OLTP database server.

The Sun SPARC Enterprise T5240 server in the application tier consumed an average of 720 Watts in a 2RU space during the execution of this benchmark. The Sun SPARC Enterprise T5240 server in the database tier consumed an average of 670 Watts in a 2RU space during the execution of this benchmark.

One Sun SPARC Enterprise T5240 server (two UltraSPARC T2 Plus chips) in the application tier demonstrated 62% better performance over the HP BL460c blade result of 2056.27 JOPS@Standard which used two 3.1 Ghz Xeon X5460 chips.

One Sun SPARC Enterprise T5240 server (two 1.4 GHz UltraSPARC T2 Plus chips) in the application tier demonstrated 2.8X better performance over the IBM p570 result of 1197.51 JOPS@Standard which used two 4.7Ghz IBM POWER6 chips.

One SPARC Enterprise Sun T5240 server (two 1.4 GHz UltraSPARC T2 Plus chips) in the application tier performed within 8% of the Dell result of 3593.58 JOPS@Standard which used two Dell PowerEdge 1950 servers with a total of four 2.8 Ghz Xeon E5440 chips. The two Dell PE1950 servers requires 2RU of rack space and consumes on average 938 Watts of power. One Sun SPARC Enterprise T5240 server has 21% better SWaP and 18% better power-performance than two Dell PowerEdge 1950 servers combined.

One SPARC Enterprise Sun T5240 server (two 1.4 GHz UltraSPARC T2 Plus chips) in the application tier demonstrated 3.8X better performance than the HP rx2660 result of 874.17 JOPS@Standard which used two 1.6 Ghz Itanium 2 chips. The Sun SPARC Enterprise T5240 server has 2.8X better SWaP and 2.9X better power-performance than the HP rx2660 server.

One Sun SPARC Enterprise T5240 server (two 1.4 GHz UltraSPARC T2 Plus chips) in the application tier demonstrated 5.1X better performance over the Dell PowerEdge 2900 result of 652.95 JOPS@Standard which used two 3.0 Xeon X5160 chips. The Sun SPARC Enterprise T5240 server has 5.7X better SWaP and 2.4X better power-performance than the Dell PowerEdge 2900.

One Sun SPARC Enterprise T5240 server (two 1.4 GHz UltraSPARC T2 Plus chips) in the application tier demonstrated 2.8X better performance over the IBM Power6 570 result of 1197.51 JOPS@Standard which used two 4.7 Ghz IBM POWER6 chips. The Sun SPARC Enterprise T5240 server has 7.6X better SWaP and 3.9X better power-performance than the IBM Power6 570.

One Sun SPARC Enterprise T5240 server (two 1.4 GHz UltraSPARC T2 Plus chips) in the database tier performed within 8% of the result of 3593.58 JOPS@Standard using the Dell PowerEdge R900 equipped with four 2.9 Ghz Xeon X7350 chips. The Sun SPARC Enterprise T5240 server has 3.1X better SWaP and 50% better power-performance than the Dell R900.

One SPARC Enterprise Sun T5240 server (two 1.4 GHz UltraSPARC T2 Plus chips) in the database tier demonstrated 5.1X better performance over the Dell PowerEdge 2900 result of 652.95 JOPS@Standard which uses two 3.0 GHz Xeon X5160 chips. The Sun SPARC Enterprise T5240 server has 5.7X better SWaP and 2.4X better power-performance than the Dell PowerEdge 2900.

One Sun SPARC Enterprise T5240 server (two 1.4 GHz UltraSPARC T2 Plus chips) in the database tier demonstrated 2.8X better performance over the IBM p5 550 result of 1197.51 JOPS@Standard which used two 2.1 GHz IBM POWER5+ chips. The Sun SPARC Enterprise T5240 server has 6.3X better SWaP and 3.2X better power-performance than the IBM Power5 550.

The SWaP metric is a measure of server efficiency ratio that includes system performance, power and space consumption on a specific benchmark. (SWaP = Perf /[ Space (RU) x Watts ] )

Power-performance is computed as watt/performance. Since power-performance is related to price/performance they are both calculated with performance in the denominator.

SPECjAppServer2004 Performance Chart as of 04/09/2008. Complete benchmark results may be found at the SPEC benchmark website http://www.spec.org. SPECjAppServer2004 JOPS@Standard (bigger is better)

  SPECjAppServer2004
JOPS@Standard		 J2EE Server DB Server
Sun 3331.31 1x Sun SPARC Enterprise T5240
16 cores, 2 chip @ 1.4 GHz US-T2 Plus
Oracle OC4J 10.1.3.3.2		 1 x Sun SPARC Enterprise T5240
16 cores, 2 chip @ 1.4 GHz US-T2 Plus
Oracle 10g DB 10.2.0.3
HP 2056.27 1x BL460c
8 cores, 2 chips @ 3.1 Xeon X5460
Oracle OC4J 10.1.3.3		 1x BL480c
8 cores, 2 chips @ 2.8 Xeon E5440
Oracle 10g DB 10.2.0.3
Sun 2000.92 1x Sun SPARC Enterprise T5220
8 cores, 1 chip @ 1.4 GHz US-T2
Oracle OC4J 10.1.3.3		 1 x Sun SPARC Enterprise T5120
8 cores, 1 chip @ 1.2 GHz US-T2
Oracle 10g DB 10.2.0.3
Rackable 1672.64 1x Rackable C2002
8 cores, 2 chips @ 2.66 Xeon X5355
Oracle OC4J 10.1.3.2		 1x Rackable S3118
8 cores, 2 chips @ 2.33 Xeon E5345
Oracle 10g DB 10.2.0.2
IBM 1197.51 1 x IBM p570
4 cores, 2 chips @ 4.7 GHz IBM POWER6
IBM WebSphere 6.1		 1 x IBM p550
4 cores, 2 chips @ 2.1 GHz IBM POWER5+
IBM DB2 v9.1
HP 874.17 1 x HP rx2660
4 cores, 2 chips @ 1.6 GHz Itanium 2
Oracle OC4J 10.1.3.2		 1 x HP rx2660
4 cores, 2 chips @ 1.6 GHz Itanium 2
Oracle 10g DB 10.2.0.2
Sybase 652.95 1 x Dell PowerEdge 2900
4 cores, 2 chips @ 3.0 GHz Xeon 5160
Sybase EAS 6.0.2		 1 x Dell PowerEdge 2900
4 cores, 2 chips @ 3.0 GHz Xeon 5160
SQLAnywhere 10.0.1
Dell 3593.58 2 x Dell PowerEdge 1950
16 cores, 4 chips @ 2.8 GHz Xeon E5440
Oracle OC4J 10.1.3.3		 1 x Dell PowerEdge R900
16 cores, 4 chips @ 2.9 GHz Xeon X7350
Oracle 10g DB 10.2.0.2

Benchmark Description

SPECjAppServer2004 (Java Application Server) is a multi-tier benchmark for measuring the performance of Java 2 Enterprise Edition (J2EE) technology-based application servers. SPECjAppServer2004 is an end-to-end application which exercises all major J2EE technologies implemented by compliant application servers as follows:

  • The web container, including servlets and JSPs
  • The EJB container
  • EJB2.0 Container Managed Persistence
  • JMS and Message Driven Beans
  • Transaction management
  • Database connectivity
Moreover, SPECjAppServer2004 also heavily exercises all parts of the underlying infrastructure that make up the application environment, including hardware, JVM software, database software, JDBC drivers, and the system network. The primary metric of the SPECjAppServer2004 benchmark is jAppServer Operations Per Second (JOPS) which is calculated by adding the metrics of the Dealership Management Application in the Dealer Domain and the Manufacturing Application in the Manufacturing Domain. There is NO price/performance metric in this benchmark.

Disclosure Statement:

SPECjAppServer2004
Sun SPARC Enterprise T5240 (16 cores, 2 chip) 3331.31 SPECjAppServer2004 JOPS@Standard.
IBM p570(4 cores, 2 chips) 1197.51 SPECjAppServer2004 JOPS@Standard.
IBM p550(4 cores, 2 chips) 1197.51 SPECjAppServer2004 JOPS@Standard.
HP BL460c (8 cores, 2 chips) 2056.27 SPECjAppServer2004 JOPS@Standard.
HP rx2660 (4 cores, 2 chips) 874.17 SPECjAppServer2004 JOPS@Standard.
Dell PE 2900 (4 cores, 2 chips) 652.95 SPECjAppServer2004 JOPS@Standard.
Dell PE 1950 (16 cores, 4 chips) 3593.68 SPECjAppServer2004 JOPS@Standard.
Dell PE R900 (4 cores, 4 chips) 3593.68 SPECjAppServer2004 JOPS@Standard.
SPEC, SPECjAppServer reg tm of Standard Performance Evaluation Corporation.
Results from www.spec.org as of 04/09/2008.

Power References:

Dell 1950 power rating estimated by applying 70% to the PSU rating reported 03/10/08 for each server:
http://www.dell.com/content/products/productdetails.aspx/pedge_r900?c=us&cs=555&l=en&s=biz
HP rx2660 power calculated as 70% of max input power reported 07/17/07:
http://h18004.www1.hp.com/products/quickspecs/12698_div/12698_div.HTML#Technical%20Specifications
Dell power rating 08/24/07 from Dell DataCenter Capacity Planner:
http://www.dell.com/content/topics/topic.aspx/global/products/pedge/topics/en/config_calculator?c=us&cs=555&l=en&s=biz
System configured with 2 x Xeon 5160 processors, 8 x 2GB DIMMs, 1 x Disk, 1 x HBA & Redundant PSU
IBM p6 570 power specifications from 80% of maximum report power consumption published here, 06/07/07, posted at
ftp://ftp.software.ibm.com/common/ssi/rep_sp/n/PSB01628USEN/PSB01628USEN.PDF
Dell R900 power estimated by applying 70% to the PSU rating, reported 03/10/08 at
http://www.dell.com/content/products/productdetails.aspx/pedge_r900?c=us&cs=555&l=en&s=biz
IBM p5 power specifications calculated by applying 70% of the power numbers published in ?Facts and Features Report?, 3/10/06, posted at
http://www-03.ibm.com/servers/eserver/pseries/hardware/factsfeatures.html

Results Summary

Certified Results 3331.31 SPECjAppServer2004 JOPS@Standard
Reference Date: Apr 9, 2008
Systems: 2 x Sun SPARC Enterprise T5240
Total Number Processors: 2,2
Processor/GHz of Server: UltraSPARC T2 Plus 1.4 GHz
Operating System: Solaris 10 8/07
Software: Oracle Application Server 10g Release 10.1.3.3.2 - Java Edition
Oracle Database Enterprise Edition Release 10.2.0.3
JVM: J2SE 6.0 update 6p

[4] Comments
Like this post? del.icio.us | furl | slashdot | technorati | digg
Posted at 08:47AM Apr 09, 2008 in Benchmarks
Tags:
Comments:

Hello,

comparing the entry for Sun SPARC Enterprise T5220

> 8 cores, 1 chip @ 1.4 <

the Sun SPARC Enterprise T5240
must be:

> 16 cores, 2 chip @ 1.4 <

Right/Wrong?

regards
Gerhard

Posted by Gerhard Krestel on April 10, 2008 at 02:58 AM PDT #

fixed thanks! I really appreciate getting notified on typos. And also make sure my internal sources of this info also fix their work!

Posted by BM Seer on April 10, 2008 at 08:07 AM PDT #

As most intelligent people know. One can make statistical results say what is wanted, especially when all of the parameters are not disclosed.
Kernel parameters, Oracle parameters, amount of RAM, and type of disks used would play an important part in benchmarking.

A lot of the comparisions are apples to oranges. Most of the time you will see better performance with 16 cores versus 4 cores. This will depend on the parameters mentioned above. I could make a 128 processor high end server perform poorly.

In the first HP comparision, Windows was the OS.
We ALL know that you can't obtain maximum performance from Windows and Xenon processors. UNIX has been around a long time for a reason.

In the second HP comparision, an entry level rx2660 server. You don't run this type of application on an entry level server. Was the OS Windows also?

I do not consider the T5420 as entry level. Isn't the Netra 240, T2000, or X4450 considered entry level servers?

Posted by puzzled.. on April 10, 2008 at 12:06 PM PDT #

Please go to www.spec.org for details as parameters are disclosed there with all required configuration details.

Remember that SPEC benchmarks are submitted by the vendors themselves, and in this hotly competitive world vendors will push what they think is the best. Core count is really bad for comparisons, *EVERYONE* implements them entirely differently, with IBM cores costing $60K/core and Sun being 1/16th that cost per core in a system.

In the first HP comparision, Windows was the OS. HP picked that to show what they thought was best.

In the second HP comparision, an entry level rx2660 server. Again HP picked this server and application.

You need to ask about T5420 pricing, I think you will be amazed of exactly how much performance you get for what price.

Also where is the HP DL580 G5 result? Where is the IBM p570 8-core result? Why aren't they publishing?

Posted by BM Seer on April 10, 2008 at 12:45 PM PDT #

Post a Comment:
Comments are closed for this entry.