Since the RAPStore (historical database) may save and maintain several years worth of market data, a redundant configuration is not practical.
If the data in the RAPStore becomes stale due to loss of the data stream or subscriber, you must manually load the RAPStore data from the RAPCache.
You may choose to install a complete secondary system for the RAPStore, as in the case of the RAPCache (the secondary instance of Event Stream Processor does not preclude this configuration), but the expense of a secondary RAPStore could be prohibitive, as there can be many terabytes of data.
If you configure a secondary system for the RAPStore, the RAPStore is loaded only by the primary instance of Event Stream Processor.
If there is a failure anywhere in the primary Event Stream Processor instance, data in the RAPStore becomes stale until manual intervention occurs. In this case, the RAPStore data is loaded from either the primary RAPCache or the secondary RAPCache, whichever is current.
Install a redundant RAPStore (an instance of a Sybase IQ multiplex server) on a separate machine. The redundant RAPStore does not run unless there is a hardware failure on the machine running the RAPStore. In the event of hardware failure of the RAPStore machine, you must, at the end of the trading day, launch the redundant RAPStore, load data from the current RAPCache into the RAPStore, and use the redundant RAPStore for the operations of the following day until the original hardware is repaired.