SPARC T3 Servers Deliver Top Performance on Oracle Communications Order and Service Management

A combination of Oracle's SPARC T3-1 and SPARC T3-2 servers running the Oracle Communications Order and Service Management application delivered world record performance with spare capacity for growth.

  • The Oracle Communications Order and Service Management application achieves a 2.4x performance boost using the combination of the SPARC T3-1 server running the WebLogic 11g application and the SPARC T3-2 server running Oracle database 11g Release 2 compared to a solution using three Dell servers (two Dell PowerEdge R610 and one Dell PowerEdge 2950).

  • Using Oracle's T-series based system solution for the application and database tiers reduces the space requirements by half compared to the equivalent Dell solution.

Performance Landscape


Tasks per second
App tier: 1 x SPARC T3-1 (1 x SPARC T3, 1.65 GHz), WebLogic Server 10.3.1
DB tier: 1 x SPARC T3-2 (2 x SPARC T3, 1.65 GHz), Oracle 11gR2 (
App tier: 1 x Sun T5440 (4 x UltraSPARC T2 Plus, 1.6 GHz), WebLogic Server 10.3.1
DB tier: 1 x Sun T5440 (4 x UltraSPARC T2 Plus, 1.6 GHz), Oracle 11gR2 (
App tier: 2 x Dell PowerEdge R610 (2 x E5530, 2.4 GHz), WebLogic
DB tier: 1 x Dell PowerEdge 2950 (2 x X5460, 3.16 GHz), Oracle DB 11gR1 (

Results and Configuration Summary

Application Tier:

1 x SPARC T3-1 server
1 x SPARC T3 processor (1.65GHz)
64 GB memory
Oracle Communications Order and Service Management 7.0.2
Oracle WebLogic 11g Middleware
Oracle Solaris 10 9/10

Database Tier:

1 x SPARC T3-2 server
2 x SPARC T3 processors (1.65GHz)
128 GB memory
1 x Sun Storage 6180 array
2 x CSM200 expansion trays
48 x 280 GB FC 15k rpm disk drivers
Oracle 11gR2 database
Oracle Solaris 10 9/10

Benchmark Description

Oracle Communications Order and Service Management is an application from the Oracle Communications Global Business Unit (CGBU) that allows Telco to handle service orders across network services and domains. It is a key component of the operations support system (OSS) suite and one suitable for T-Series servers. The key metric is tasks per second (tps) demonstrating the ability to handle a large number of orders.

The benchmark was developed by the Oracle Communications Order and Service Management (OSM) team. Results have not been shared externally and are mostly being used for sizing purposes. This particular workload, known as the "Classic" model, simulates local fulfillment of OSM orders. An order is a unit of work that separates out into a varying number of tasks. The amount of completed tasks per second forms the basis of the metric. In a typical deployment scenario, OSM takes its input from the CRM system (e.g. Siebel), breaks down the order into a set of tasks, and channels these tasks into the service activation system of a Telco operator.

The workload stresses the database backend with a lighter load on the application tier. As a guideline, a well-balanced design specifies 2-3x the compute power on the database tier over the application tier.

Key Points and Best Practices

  • The average server utilization for the database server (SPARC T3-2) was 70%. The average server utilization for the application server (SPARC T3-1) was 50%.

  • The SPARC T3-2 and SPARC T3-1 servers require a total of 5U rack space. The Dell solution required a total of 4U rack space. Including in the difference in performance, the Oracle T-series solution provides much better space/performance solution (much more performance in similar space).

See Also

Disclosure Statement

Copyright 2010, Oracle and/or its affiliates. All rights reserved. Oracle and Java are registered trademarks of Oracle and/or its affiliates. Other names may be trademarks of their respective owners. Results as of 9/20/2010.


Any word on any revisions to the processor core factor table that Oracle has published?

Right now, the T3 falls under the 'other' category at 1.0 cpu licenses per core. So that SPARC T3-2 would require 32 cpu licenses@40k per license for a cost of $1,280,000. If they give it the same core factor as the T2+ then the db cost would drop to $640,000.

Compared to the $160,000 cost for the Dell 2950 (and why did they use an old Dell w/ 11gR1 vs a newer R710 w/ the x5677 cpu's and 11gR2).

Same issue on the App server tier, you given the loading delta and 2-3x rule cited. A single R710 w/ two 6 or 8 core cpu's would probably do the trick and keep the core count down to help with the software license costs (which is huge when you start talking about anything oracle and can often force a customers decision one way or the other)

As for the space issue, a pair of Dell R810's or R815's would give you over 2x the performance in the same amount of rack space. This would put it on par w/ the T3 based system at a fraction of the cost for the Oracle license. Also missing is the disk specs for the other two configs. I'd assume that they were similar and if not then the question is what you are testing: the server or the disk. It would also be interesting to see the impact of an F5100 on the db results.

Posted by John Garner on September 21, 2010 at 03:57 AM PDT #

Post a Comment:
Comments are closed for this entry.

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

« July 2016