Data Fusion - High Performance Clustering and In-Memory Caching
Data Fusion - integrating complex data from multiple sources
High Performance Clustering and In-Memory Caching
Clustering reduces the number of physical I/Os needed to access a group of related objects. The application designer can cluster objects that are generally accessed together into a single logical “page”. Reducing the number of I/Os needed to service a transaction can dramatically shorten transaction times and increase overall system throughput.
Caching reduces I/Os by attempting to keep frequently used data in RAM; or by reusing RAM immediately if data is being streamed to or from disk. Frequently accessed data, such as dictionaries or glossaries, may be locked in memory. Incoming video (or other binary data) may be streamed to disk through a small RAM cache.
In Objectivity/DB the clustering and caching strategies are defined at runtime on a per-thread basis, unlike RDBMSs that rely on a database administrator to control server behavior.
