The survivability of OS is very important for the whole system because OS is the base of information system or network system. Based on the analysis of resources, services and functions of the OS, this paper proposed the concept of a integrity running environment (IRE) owing to the particularity of the OS survivability, and then, puts forward the new definition, namely the OS survivability is that the OS will still be able to provide a IRE to critical processes when an exception occurs. Under the guidelines of the new definition, redundancy, isolation, decentralization, functional complexity and exception sensing as the survivability strategies for improving the OS survivability are discussed. Analyzed the technologies of supporting the strategies, it is pointed out that architecture combined the microkernel and component technology is very conducive to implementation of improvement of the OS survival.