Massive Databases

MVX private cloud technology creates a large distributed memory service
- a Memory Cloud - from RAM in existing servers



MVX accelerated the customer's back testing by up to 20X with
no changes to the application

Modern life is increasingly touched by giant data collections:

• Libraries of video content
• Movie scenes generated from high resolution texture maps
• Raw sensor data, ranging from surveillance cameras satellite images to seismic data
• Flight, shipping, and reservation information
• Business analytics, often using demographic and marketing data
• Life sciences and genomic information

These data sets each quickly add up to terabytes or even petabytes -- much too large to fit into even large-scale local RAM on a server. These growing data sets create huge performance slowdowns in applications where latency and throughput matter. Moving data from storage systems and over networks takes time, making multicore processors wait for data they can’t get fast enough. Currently available solutions are inefficient workarounds that don’t solve the problem.

To derive benefits from these large data sets requires processing such as analysis, modeling, or transformation. The more detailed or frequent the processing, the more accurate, applicable and valuable the results. The central issue facing large databases is, how can we get this data around the data center, from where the data stored to where it needs to be processed?

RNA MVX is a breakthrough in making large active data sets much more accessible across the data center. A Memory Cloud allows large-scale memory caches to be mobilized -- multiple terabytes in size. With the most frequently accessed data records in RAM, data base operations are dramatically accelerated.

The effect is particularly powerful when database operations match two large databases against each other, such as a business with thousands of customer records and millions of sales records, or geographic data sets with thousands of locations matched against millions of data points. With MVX, a large Memory Store can hold the smaller data dimension entirely in memory while the larger dimension is optimized with MVX’s Memory Cache. By spreading the data across many servers’ RAM, data access avoids both storage and network bottlenecks for huge improvements in performance.

Fast processing of massive data sets is becoming a key differentiator in many industries. Faster response to events, better customer service, and real-time efficiency all benefit from accelerated access to large amounts of data. RNA MVX provides a central technology for achieving this aim.