In-Memory Database Applications

From its inception, SGI has been about one thing: BIG DATA. Whether crunching it, storing it, or visualizing it, SGI systems have always excelled at helping people turn mountains of data into knowledge. So it is no coincidence that SGI systems are extremely well suited to running the data management software people use to manage overwhelming amounts of data. SGI platforms based on industry standards offer outstanding performance and total cost of ownership for users with demanding database challenges.

Designed to efficiently handle massive datasets and complex computational problems by enabling unsurpassed scalability, SGI UV is uniquely suited to hosting in-memory database applications. The SGI® UV server family is based on SGI® NUMAlink® shared-memory architecture. This architecture provides unmatched scalability - giving users the ability to expand memory, I/O and processors independently to efficiently address specific processing requirements.

  • Memory Scalability: A single UV node, running one copy of the Linux® operating system, can scale up to 16TB of shared memory respectively using a single system image of the operating system. Effectively, SGI UV servers can accommodate as much of the database in memory as the software itself will permit.
  • I/O Bandwidth: SGI UV servers have demonstrated sustained I/O bandwidth of over 7 gigabytes per second and achieved a world-record 1 terabyte per second result on the STREAM Triad memory bandwidth benchmark.
  • Processor Scalability: SGI UV servers support from 1 to thousands of cores per system, all operating against a pool of globally shared memory.

Because of its unique shared-memory scalability, SGI UV delivers efficient management of large data sets, access to massive data sets, and execution of more instructions per cycle, all of which boost overall performance of in-memory database applications.

SGI In-Memory Database solutions are enhanced by SGI® InfiniteStorage products, which are designed to ensure that the storage performance is high enough to keep up with the server architecture. Accelerator technology can provide substantial performance improvement for database applications that spend a majority of their run time on a set of specific algorithms. Examples include text search and pattern recognition routines, data mining, and very large database solutions that require sub-second response times.