A software based online memory test for highly available systems

In this paper we describe a software based in-system memory test, capable of testing system memory in both offline and online environments. A technique to transparently "steal" a chunk of memory from the system for running tests and then inserting it back for normal application use is proposed. Implementation of the proposed methodology can significantly improve the system's ability to proactively detect and manage functional faults in memory. The solution does not impose any hardware requirements and therefore lends itself for easy deployment on all kinds of systems. An extension of the methodology described is expected to be applicable for in-system testing of other system components as well.