Hardware appliance technologies have become a common solution as the complexity of computing environments grows significantly with each server needing to be integrated and tuned. Integration costs are rising to the point that the ongoing lifecycle cost of a given component outweighs its acquisition cost. An appliance is not built for general purpose computing, but instead provides consistently good price performance and supportability (low TCO) for a specific type of workload. The ability to select and tune the hardware, software, and OS for a specific purpose allows ultimately greater and more consistent performance. One reason to produce an appliance instead of a software offering is the inherent ability to maintain performance, reliability, and compatibility within the environment. In a non-appliance environment, all of these burdens would be placed on the IT staff and end users.
Quasar provides the capability to effectively use memory and robust parallel processing to address the scaling and performance requirements of very large databases.
Quasar is built on a FPGA core that uses dedicated-circuit speeds to accelerate the indexing, indexing management and ingest/query functions of the database it is accelerating.
Quasar also utilizes memory to surround its FPGAs in order to minimize system latency which then results in nearly theoretical maximum speeds.
Quasar can scale to fit the size of your data situation. Unlike other options of adding additional servers, Quasar provides for linear growth, or more over if you adding two-times the number of Quasar units this will double your capacity. Most, if not all, other hardware options will reach a point of diminishing returns, which is the point when adding more hardware fails to increase capacity or speed.
NexQL’s Quasar provides Intelligent Index Acceleration as an adjunct processing system for database systems. Quasar works with Oracle and other databases to manage very large or complex indexes reducing latency and I/O while supporting analytic functions. Quasar’s patented design uses state of the art technology implementing customizable logic within Field Programmable Gate Arrays (FPGAs). These very fast processors with a wide 128 bit memory data bus, twice as wide as most current computers, and 16 gigabytes of Random Access Memory are housed on the Quasar Index Engine. Each Quasar appliance can support up to 14 Index Engines.
Quasar is a highly parallel processing system that may be configured with thousands of Index Engine blades that provide a very high degree of parallel processing devoted to searching of complex index structures.
A major advantage of the highly parallel Quasar architecture is its ability to scale to support extremely large indexes. Indexes of hundreds of billions of records can be deployed, providing near linear response times. Quasar’s highly scalable architecture provides both very high ingest and search capacity for very large database installations. The Quasar indexing system is dynamic and self-optimizing reducing the on-going need for standard index maintenance and the need for additional DBA staffing.
NexQL has developed Index Processors that use large memories for storage of index data. Each processor can perform atomic insert, update, delete and scan operations on any index within its memory. Even for a very large index segment of 100 million keys these operations can finish in tens of microseconds.
Quasar uniquely addresses Intelligent Index Acceleration as a solution for database performance enhancement by supporting application specific indexing through the FPGAs. Custom analytics, too complicated for relational databases, can now take advantage of Quasar’s FPGAs additional performance enhancements. Indexes such as Geo Spatial or multi valued indexes can be processed and searched at tremendous speeds.
The flexibility of the Quasar system can provide complex relationship mapping between entities. Graphs of entities constructed with multiple relationships goes far beyond the traditional relational model.