The primary goal of TruData is to eliminate the requirement of sharding your database once it has reached a certain size; avoiding the headache of additional infrastructure and unnecessary division of a dataset in order to scale. Tables can now grow to hundreds of billions of rows and continue to run directly off disk with no performance degradation. In addition, TruData utilizes custom data structures to automatically optimize and normalize data structures as necessary. Using this approach, TruData is capable of providing support for multiple secondary indexes and a high compression rate of 25-50x while avoiding the read and write performance penalties found in other traditional database systems.
TruData’s data structures are relational in nature. With proprietary automated scripts, data warehousing for backup or offline archival purposes is simple and fast. TruData data structures are designed to allow data to migrate between hardware optimized for analytics and warehousing with minimal restructuring. This not only allows fine tuned control over compression vs. performance, but specifically minimizes the amount of time for data to be recategorized as high-priority analytical or offline archival. TruData enables trillions of entries to be analyzed, stored and archived, all on the same physical server. Splitting datasets over multiple machines is no longer a time consuming requirement, but instead an optional endeavor in infrastructure optimization.
While TruData utilizes a proprietary data structure format; it is designed to work with traditional analytical systems on the market, such as NoSQL solutions or Hadoop. By default, all TruData data is indexed to provide primary key or date and time sequential blocks of data, allowing extremely fast exports with no additional work. As a result, in the case of extremely high grade critical analytics, any TruData data can be easily and quickly exported into a third party solution where web scale or cluster sharding for performance is necessary.