Note Check the solidDB Release Notes for any limitations that are associated with using a grid in the current release.
One of the main benefits of using a grid is the opportunity to increase system throughput while retaining the low lookup latencies of a standalone solidDB server. However, the throughput and latency vary depending on the type of operation.
▪ For single row lookup operations, the throughput is likely to scale linearly as the number of nodes grows while latencies stay comparable to those of a standalone solidDB server.
Note Latency includes the network latency unless the client application runs in same hardware as the grid node.
▪ For single row write operations, the throughput is not likely to scale linearly with number of nodes since the row must be written to multiple nodes to allow redundancy in case of failure. The theoretical maximum throughput with N nodes and a replication factor M is N/M times the standalone write performance. The practical throughput improvement will not be quite as high with a large number of nodes in the grid.
The latencies for single row write operations are likely to be similar to those of a solidDB standalone server.
Note Latency includes the network latency unless the client application runs in same hardware as the grid node.
▪ For complex read and write operations that involve multiple rows in different grid nodes, the grid is not likely to provide performance benefits. Generally, latencies are expected to be longer because the rows must be retrieved across the network. For some operations with a large grid, the throughput might be higher than with a solidDB standalone server.