Sybase MPP shows excellent scalability to 128 CPUs
Posted in 1995
SYBASE ACHIEVES NEAR-PERFECT SCALABILITY ON ONE OF THE WORLDS LARGEST COMPUTERS: 128 PROCESSOR IBM/RS6000 SP Sybase MPP Demonstrates Unlimited Scalability for the Worlds Most Demanding Data Warehouse and Massive Decision Support Systems EMERYVILLE, CA, November 7, 1995 -- Sybase, Inc., The Enterprise Client/Server Company, today announced results of a test conducted at the Maui High Performance Computing Center (MHPCC). Sybase MPP (previously called Sybase Navigation Server) demonstrated near-perfect scalability on an IBM RS/6000 Scalable POWERparallel (SP) system using 128 high performance RISC processors. Sybase MPP is Sybases scalable, open parallel database for unlimited scalability for large decision support applications. The demonstration consisted of a credit-card database containing a 100 million row table distributed across all of the processors. A Powerbuilder graphical interface displayed 12 different queries running against the 100 million-row database. A variety of database queries were performed, including a variety of complex queries and traditional credit card queries for gathering information on account balances, summaries by customer, and summaries by region. Sybase MPP demonstrated near-linear scalability from one to 16, 64, and 128 nodes, and scalability of over 99 percent across the range of 128 processors. Additionally, the demonstration included table scans, two- and three-table joins, as well as parallel inserts and deletes. For example, a complex sum of all 100 million rows in the database was achieved in 94 seconds, and a three-table join with an account summary by customer was achieved in under 20 seconds. The MHPCC is a test bed for scalable parallel technologies, said Margaret Lewis, associate director of marketing for MHPCC. This demonstration allowed us to showcase the outstanding scalability of Sybase MPP on 128 processors of our IBM RS/6000 SP system. Not only did Sybase MPP scale well, it balanced the workload efficiently over the processors. Customers who need to solve large scale data mining and data warehousing problems will be very interested in Sybase MPP. Sybase MPP is the first and only production open systems DBMS which provides near-linear scalability and parallelism on all database operations, said Dennis McEvoy, vice president of Sybase Enterprise Products. Unlike other parallel products on the market, Sybase MPP has been designed for a no-contention parallel environment and near-perfect scalability to support terabytes of data on hundreds of processors. This demonstration on the IBM RS/6000 SP demonstrates the stability and portability of Sybase MPP. We see a large demand for Sybase MPP in Japan and were pleased to sponsor this event, said Yoshimi Ogawa, President of Sybase K.K. We conducted this demonstration for some of our largest customers, many of which need the performance and scalability of Sybase MPP for their data warehouse applications. * Sybase MPP Sybase MPP (formerly called Navigation Server), a member of the Sybase System 11 family of database products, is the only portable parallel database product with a shared-nothing architecture for solving the largest database problems in decision support, data mining, and data warehousing. It is uniquely architected to distribute large databases across dozens to hundreds of independent processors, allowing all database operations to be performed in parallel. Sybase MPP includes parallel database queries, stored procedures, inserts, deletes, and complex table joins. These capabilities enable Sybase MPP to scale with near-linear performance on a variety of computing platforms. * Availability & Pricing Sybase MPP is in production today on the AT&T 3000 series of systems. A beta version for the IBM RS/6000 SP is available today, and beta versions for Hewlett-Packards HP 9000 family of business servers, and clusters of Sun Microsystems SPARCcluster 1000/2000 PDB (parallel database) are currently schedule to be available this quarter. Pricing for Sybase MPP, which is based on the processor configuration supported, starts at $110,000. * Maui High Performance Computing Center (MHPCC) The MHPCC was developed and is managed by a consortium led by the University of New Mexico under a cooperative agreement administered by Phillips Laboratory, United States Air Force. It houses the worlds second largest installation of the IBM RS/6000 SP systems, with a total of 480 processors, offering a capacity of 125 billion calculations per second (125 gigaflops). The MHPCC supports a broad customer base, including Department of Defense, commercial and academic users. Together with Sybase, MHPCC is offering leading-edge technology, such as parallel databases, to its customers. Both companies are working closely together to offer a broad range of services to commercial and government users, including education, benchmarking, and testing of large parallel databases for massive decision support applications. For additional information on the MHPCC and its services, call (808) 879-5077 or refer to the web site http://www.mhpcc.edu. Sybase, Inc. Sybase, Inc. is a worldwide leader in client/server software. Sybase and its wholly owned subsidiary Powersoft Corporation are focused on four major market segments: on-line transaction processing, data warehousing, mass deployment and electronic commerce. Sybase's software products -- database, middleware and tools -- provide customers with desktop to enterprise solutions. The company's mission is to provide its customers with an open, adaptable information systems architecture that enables them to develop and deliver complete information systems solutions to facilitate rapid business change. Headquartered in Emeryville, CA, Sybase is the sixth largest independent software company in the world. # # # SYBASE, System 11, SQL Server, Sybase MPP, Navigation Server, Powersoft, and The Enterprise Client/Server Company are trademarks of Sybase, Inc. All other company and product names may be trademarks of the respective companies with which they are associated.