A set of solidDB servers that run in multiple connected nodes and appear to client applications as a single logical database.

A solidDB server that is configured to be part of a grid and has built-in table partitioning and replication support.

The grid node where a statement is executed or co-ordinated. The co-ordinator node can send query fragments to remote nodes and collect results.

Nodes, other than the coordinator node, that are part of the grid.

Part of a table, also known as a shard.

Table that is divided into multiple partitions. Each partition contains a subset of the table rows and is stored on a number of grid nodes. The number of nodes that store a replica copy of each partition (a replication unit) is determined by the replication factor, see Replication factor.

Table that consists of a single partition. Every node has a replica copy of the whole table.

A replica copy of a table partition. Replica copies of the same table partition form a replication group.

Replication unit that stores the primary instance of rows from a partition of a partitioned table, or the primary instance of a non-partitioned table. The primary replication unit is sometimes referred to as the ‘primary partition’. All updates to the row (or table) are performed in the primary replication unit. The node that contains the primary replication unit is determined by the grid, see Partitioning key.

Replication unit that stores replica copies (secondary instances) of rows from a partition of a partitioned table, or a replica copy of a non-partitioned table. The secondary replication unit is sometimes referred to as the ‘secondary partition’. Data can be read from but not written to a secondary replication unit. There can be multiple secondary replication units; the number is determined by the replication factor.

Contains the primary and secondary replication units for a single table partition. Each replication unit in a replication group is stored on a separate grid node.

Column or columns that uniquely identify a partition. Each row in a partitioned table includes a partitioning key that is used by the distribution formula to identify the replica group to which the row belongs. See Distribution formula.

Defines required number of instances (n) of a partitioned table row or a non-partitioned table that the grid must maintain (one in the primary replication unit and one in each of n-1 secondary replication units).

Determines the replication group to which a row in a partitioned table is allocated.

An SQL statement that accesses data from multiple nodes.

Maintains the high availability of the grid. Automatically runs in the leader node, monitors replication status between nodes and responsiveness of grid, and fixes issues.

A software component that monitors the grid nodes and attempts to restart any nodes that fail.

See solidDB consensus algorithm for definition of terms related to the solidDB consensus algorithm.

Holds the table partitioning and location information for all tables (both partitioned and non-partitioned) in the grid. The data partitioning map is an internal data structure that is distributed to all drivers and nodes so that SQL statements can locate the data. Central to the operation of the grid, the data partitioning map is used by the replication process to decide which nodes receive copies of a row of data.

For example, the network addresses of all nodes in the grid, and the data partitioning map that is stored in system tables.