Difference between revisions of "About AVA"

Latest revision as of 14:19, 13 May 2013

AVA Database is the next generation database designed in the Big Data era. It is aimed at the goals that traditional databases fails to achieve. AVA embraces the design features and technologies that specifically target at the speed and scalability. It can run queries 10x - 100x faster than traditional databases, achieve near real-time event processing, distribute data storage across the network, and run in massively parallel fashion.

AVA Database suites come with different versions to custom-fit your IT requirements. Please contact our sales representatives for further information.

Main Advantages

• Run 10x - 100x Faster Than Traditional Databases
  • In traditional databases, such as Oracle or MySQL, executing queries that contain either "group by" clause or requires joins ( both inner and outer ) are notoriously time-consuming. AVA provides super fast algorithms to deliver fast queries. For example, joining two tables with 100 million unique keys ( compound keys with more than 3 columns involved ) takes only less than 1 minute, sometimes only about 10-15 seconds.
  • Some queries are either not available or impossible to execute given the prohibitively extensive and time-consuming operations. For example, the time series operations to calculate statistics on a moving window. It only takes a fraction of a minute or seconds to process on a large set of data ( > 100 million records ). AVA's suite of scientific analytics and data mining tools provide the speed even unthinkable to existing commercial databases.

• Near Real-time Event Processing

Thanks to high performance of the database, event processing and analytics are executed almost real time, with extremely low latency. Streaming of information ( messages, stock quotes and news ) are processed at the speed of 10-50 million records/second on a single thread.

• High Compression Ratio

With the Column-based storage design, AVA is able to compress the data to at least 30% (3:1 compression ratio) of the original size, with extreme cases of 10:1 ratio. AVA’s proprietary compression ratio can perform high speed inline compression. It is also able to save compressed to storage devices, greatly enhance the system I/O performance and reduce costs.

• Distributed Storage

AVA can store a single instance of database or data table to multiple sectors within the same storage device as well as across the entire network within an enterprise. It is flexible enough to work on virtually any IT infrastructure, and it can utilize the existing IT resources with minimum reconfiguration.

• Optimized Storage

AVA provides several storage solutions, and has fully optimized its disk I/O. It can load/save data tables of multiple terabytes to petabytes with linear growth in the time spent. It also utilizes the high compression ratio and saves the data tables 70%-90% smaller than the original size.

• High Scalability

AVA’s Enterprise Edition can take the database to over Petabytes without much performance penalty. The Enterprise Edition is a multi-thread, multi-process version which scales linearly over multiple commodity machines and storage. It also employs the the industry-leading MapReduce framework with Share-Nothing architecture.

Main Design Features

• In Memory : the entire database is in-memory when launched. It can be saved to disk for eventual persistence.
• Column-based : the memory model is by column, rather by record. The Columnar model greatly enhances the performance of certain database applications.
• In-line compression/decompression : inline, super fast compression/decompression. Also support the same compression on disk.
• Optimized Storage I/O : disk I/O is fully optimized to minimize the I/O penalty.
• Distributed Storage : the database can be saved to arbitrary number of storage device across the same network.
• Massively Parallel : multi-process model across multiple machines with minimum memory-sharing and achieve linear scaling.