We are not kidding.
SPARC M7/T7 results are now coming in from our partners and the results are truly worth noting.
Here is one more example to add to the list.
Hangzhou New Century Electronic Technology Co. Ltd. (NCI), located in Hangzhou, China, is an Oracle partner specializing in systems integration, hardware and software distribution, and technical support services for the tobacco industry. They develop and distribute an integrated management and control application called "The Tobacco Business Operation & Maintenance Platform v4.0", which is a widely recognized solution based on Oracle’s Java EE architecture. This platform provides the tobacco industry with a unified platform for operations,
management and safety, and includes cloud compute & big data
As the story has been going with the SPARC T7, NCI heard about the highly acclaimed analyst reviews, 20 new world record benchmarks, the numerous customer quotes (B&H Photo Video: "run 83x faster on SPARC T7"), and decided to try out the powerful T7-2 server on the Tobacco Business Operations & Maintenance Platform. Their objective was to run tests to measure the performance results comparing the Oracle Database 12c with the IBM DB2 Database, both running on the Oracle SPARC T7-2 servers.
The hardware configuration included two Oracle SPARC T7-2 servers with 512GB of memory to deploy Oracle Database 12c RAC, and two Oracle SPARC T7-2 servers with 512GB of memory for deployment of Oracle Weblogic 12c Cluster. A single Oracle FS1 Flash Storage System was used to deploy Oracle RAC ASM, and for operation and maintenance of data storage.
The software configuration included Oracle Solaris 11 as the operating system; Oracle Database version 220.127.116.11.0; Oracle WebLogic version 12.1.1 for the middleware; and IBM DB2 Database version 10.5.0.3.
Test Results: 8 minutes to 7 seconds!
Running the same series of SQL statements on the IBM DB2 database took just over 8 "minutes",
when the same exact run on the same hardware took just under 7 "seconds" with Oracle Database 12c.
The tests were run with the Oracle Database In-Memory option enabled and utilized the SPARC T7 Data Analytics Accelerator (DAX) co-processors. These
results were compared to the exact same hardware configuration running
the IBM DB2 Database. The test database file size in both cases
was 256MB. This was a test to see if the Oracle on Oracle solution
could demonstrate a significant advantage.
NCI realized two primary factors were responsible for that dramatic advantage of the Oracle on Oracle solution:
1) Data Analytics Accelerators (DAX) are individual co-processors on the SPARC T7 chip which perform query-related operations directly through the hardware, thereby improving Oracle Database performance significantly. NCI was able to utilize DAX hardware acceleration for the Oracle Database 12c in-memory database operations on this test, thereby relieving the SPARC T7 cores from this analytics workload.
2) Oracle Database In-Memory is a new option with Oracle Database 12c which works with DAX on the SPARC T7 chip to transparently accelerate analytic queries by orders of magnitude. Oracle Database In-Memory optimizes both analytics and online transaction processing workloads (OLTP), delivering outstanding performance for transaction processing while simultaneously supporting real-time analytics for business intelligence.
A couple of notes on this test:
1) The SPARC M7 DAX functionality is used by default when Oracle Database In-Memory Query Accelerator is enabled. No additional administration is required to turn this functionality on.
2) The database file size of 256MB is a very small size and not applicable to most use cases. A larger database would yield even better performance results.
3) Oracle offers a tool called the In-Memory Advisor which could be used here for assessment of the workload and recommendations regarding how to optimize the Oracle Database performance. See the Oracle Database In-Memory Advisor White Paper for details.
For more information, details and system sizing help you can contact the team via firstname.lastname@example.org