AT ORACLE: News
SPARC Torches BenchmarkBy Diana Reichardt
Oracle’s new SPARC Supercluster, Oracle Exalogic Elastic Cloud T3-1B, and Oracle Solaris 11 break records and set new standards for performance and availability.
Oracle has achieved a new world record on the industry-standard TPC-C benchmark for online transaction processing (OLTP) database performance using a new SPARC server. Oracle achieved 30.2 million transactions per minute, beating IBM’s best performance by three to one and HP’s best performance by more than seven to one.
Oracle CEO Larry Ellison and Oracle Executive Vice President of Systems John Fowler took the stage in Santa Clara, California, in December to tout the new world record and reveal details of Oracle’s new SPARC Supercluster family of products, Oracle Exalogic Elastic Cloud T3-1B, and Oracle Solaris 11.
The TPC-C benchmark results were achieved on Oracle’s new SPARC Supercluster with SPARC T3-4 servers running Oracle Database 11g on Oracle Real Application Clusters. The SPARC Supercluster configuration consisted of 27 SPARC T3-4 servers, 138 Sun Storage F5100 Flash Array storage devices, and 97 Sun Fire X4270 M2 high-capacity storage servers that delivered massive data capacity and accelerated I/O performance.
“This is with a legit Oracle database with 1 quadrillion rows of data,” Ellison told the audience. “I haven’t said the word quadrillion since I was in grammar school learning numbers. I just have never had the need to use that word in computing. This is 43 trillion transactions per day, with an average response time of less than half a second.”
SPARC Supercluster Family of Servers
After congratulating the Oracle team responsible for the record-setting system, Ellison unveiled Oracle’s new SPARC Supercluster family of products, which combine new SPARC T3-4 servers with Sun FlashFire, InfiniBand QDR networking, Oracle Solaris, and the Sun ZFS Storage Appliance in a fault-tolerant configuration.
The new products integrate servers with InfiniBand, flash, memory, storage, and all the hardware and software you need to run your business today, Ellison said.
Oracle’s SPARC Superclusters are expected to be available in 2011 and will include SPARC T3-2, SPARC T3-4, or Sun SPARC Enterprise M5000 servers; Sun ZFS Storage 7420 appliances; Sun Network ZFS InfiniBand Gateway Switches; and Oracle Gold support.
“By engineering both the software and the hardware to work together, we can deliver world-record performance, the lowest-possible TCO, the best reliability, and the best security—and that’s what Oracle Sun is all about,” Ellison said.
Oracle Exalogic Elastic Cloud, SPARC Version
Ellison also introduced Oracle Exalogic Elastic Cloud T3-1B, an integrated hardware and software system based on SPARC and Oracle Solaris and tested and tuned by Oracle to run Java and non-Java applications with extreme performance. This dedicated middleware machine is built on a SPARC T3-1 server and delivers that system’s scalability, security, and integrated chip features.
Oracle Exalogic Elastic Cloud T3-1B combines SPARC servers running Oracle Solaris 11 Express with InfiniBand-based I/O fabric, Oracle WebLogic Server, and other enterprise Java-based Oracle middleware products.
“This thing is just tuned to make Java scream,” Ellison said. “So if you’re running Oracle Fusion Middleware, this is a fabulous machine. The SPARC version of Oracle Exalogic Elastic Cloud runs Java faster than any other computer in the world.”
Oracle Solaris 11
Fowler took the stage following Ellison to provide details on Oracle Solaris 11, which is due out in 2011.
“Hardware without software is a doorstop, and the core to the SPARC platform is Oracle Solaris,” Fowler said. “For more than a decade, performance, availability, security, and management have been the cornerstones to Solaris. We are raising the bar again with Oracle Solaris 11. It is a complete reworking of the enterprise operating system to give you next-generation networking for scale and performance to get to the kinds of cores, threads, and memory that people are going to see in hardware right around the corner.”
Fowler said that measuring downtime in seconds and uptime in years has been a mantra for Solaris for a very long time. Oracle Solaris 11 takes availability to a new level, he said, by changing the boot architecture for boot times of a few seconds; changing system patching and updating to eliminate boots; and improving fault management, reliability, and service management technology.
“Throughout Oracle Solaris 11, we have improved these capabilities of availability, designed for mission-critical enterprise applications,” he said. “This work is also being integrated with upper levels of the stack like middleware and database so that together we have extremely fast and resilient high-reliability capabilities across the entire runtime.”
Wrapping up the Santa Clara event, Fowler explained that Oracle wants to save customers the time and money they spend integrating storage servers, operating systems, and other technologies by delivering ready-to-run systems that provide the absolute-highest performance and technical value. “The SPARC Exalogic [Oracle Exalogic Elastic Cloud T3-1B] and the SPARC Supercluster [machines] are outstanding examples of what I expect to be a very exciting future in delivering completely engineered systems for enterprise mission-critical computing,” he said.