I wrote about GridSQL and EnterpriseDB a while back. EnterpriseDB bought GridSQL (from Mason who now works at EnterpriseDB) and has been working to improve it. Yesterday, they announced that they are open sourcing it.
GridSQL is really for data warehousing but can be applied to some OLTP/reporting combination databases. What it does is take a query and spread it out to run in parallel against a set of cheap PCs. There is a controller machine, called a coordinator, that keeps track of what is running and where.
Because there is a central point of failure (the coordinator), it is not a high availability solution. But, also because of the coordinator, you can hot swap, add or remove nodes while users are accessing the data. Very cool technology.
LewisC
gridsql enterprisedb
