Omnidex Grids distribute data across multiple nodes. Nodes can be evenly distributed, or they can be distributed based on criteria, such as company, geography or time. Omnidex Grids are ideal for very large databases as they distribute the data across multiple servers and allow scalability through parallel processing. At the same time, Omnidex Grids create new options for updating the database.
Omnidex Grids do not directly support online updates, meaning that applications cannot issue INSERT, DELETE or UPDATE statements to an Omnidex Grid. Instead, applications can add new nodes to a grid to quickly integrate new data into the database. A new node can be independently created and indexed before it is formerly added to an Omnidex Grid. Once it is ready, it can be added to the Omnidex Grids, allowing applications to access the new data.
This approach requires attention to the partitioning scheme for the grid. If all the rows to be added can be isolated into a single partition, then only one partition has to be reindexed; however, if many nodes are affected, then all will have to be reindexed. For example, if an Omnidex Grid is partitioned by time, then a new node can be added that reflects the most recent additions. Conversely, if an Omnidex Grid is partitioned by geography, then it is likely that all nodes would be affected when new rows are added. The same principle applies to database updates and deletes as well.
See also: