Performance results for real-time databases

Strictly speaking, a real-time databases is one whose response time for certain transactions can be shown to meet temporal constraints, commonly expressed as a worst-time requirement.

Normally a database is designed to return data as fast as it can, without any hard constraint on how quickly it does so. This is in good part because the response time is highly dependent on the structure of the data stored. The same query with a different value for a key could vary by orders of magnitude depending on the number of rows with the key value in the query. Because of this, a true real-time database is a combination of a real-time operating system, a well defined and constrained execution environment, a real-time database management system, a well-defined database, and a well-defined set of transactions. In practice, such a tightly defined and constrained system is rare, and data gathered for one are not applicable to other such systems, since each is highly idiosyncratic. As a result, data, when available, tend to be comparative or theoretical.

In-memory database management

The closest thing to a real time database is an in-memory database. This is usually done in one of three ways. Some data comparing performance for the last two alternatives can be found in:

Additional sources of performance data Below are references to papers with performance data for high-performance databases.