Patent-pending index structures, that contain inherent information, minimize the need to read from disk, eliminating a common database bottleneck. There are two parts to Fetchless Operations - Combinator Indexes and Skimmed Aggregations. Each of these features contributes to the unrivaled analytical performance that DexterityDB provides.
Aggregations and statistics are computed using inherent information indirectly stored in indexes, eliminating fetches from disk. This extra information can be used to improve the performance of summations and averages, as well as many other statistical functions that are common in report generation and application analytics.
Unique data structures eliminate the key duplication that is common in database storage. Eliminating these redundant processes leads to quicker ingestion and more efficient index storage. This can be extremely useful when it comes to using DexterityDB as a supplement to existing OLTP configurations. Ingress is very fast, allowing DexterityDB to keep up with live changes on the transactional database and to be used effectively in ETL processes.
Patent-pending index structures accelerate logical intersections (AND & OR) by eliminating the need for data fetches from disk. DexterityDB's indexes can be directly compared without locating the actual data that they refer to. This means that filtering and comparison operations, which are commonly used in searching and analytical queries, are extremely quick.
Intelligent local sharding allows for reduced locking, multithreaded query execution, and reduced storage needs. Data is automatically divided up without user input in order to schedule and optimize complex database functions based on hardware characteristics of the server, leading to quicker retrieval of relevant information to satisfy queries.
A Document-based store that can act like a Row Store, or even a Column Store, depending on what database it is plugged into and how it is used. At its base, the DexterityDB engine was designed as a Document-based Store. This allows it to be flexible. When plugged into various row-based databases, it acts as a Row Store and has the insert and raw data retrieval benefits associated with Row Store engines, but has characteristics of a Column Store, giving it great analytical performance.