This guest editorial by MontaVista Software's VP of Engineering, Kevin Morgan, makes the case for obtaining hard real-time performance from Linux through a kernel preemptibility enhancement, rather than by using the approach of adding a second kernel such as RTLinux or RTAI. Morgan writes . . .

“Real-time is a term that characterizes a particular application. Hard real-time means an application fails catastrophically if deadline requirements are not met. Soft real-time means an application suffers degradation in quality, but not catastrophic failure, if deadline requirements are not met. Both hard and soft real-time are terms that are clock time independent. A hard real-time requirement might be relatively slow (i.e., if an event is not serviced within one minute, an engine overheats and breaks down). A soft real-time requirement might be relatively fast (i.e., if a buffer on a networking card is not drained within 200 microseconds of an interrupt event it will overflow, but it will recover from the loss of data). The important point is that soft and hard real-time are attributes of the impact of failure to meet deadlines, and the specific timing requirements are independent of the “hardness” of the requirement.”

“Linux is capable of meeting a wide variety of real-time requirements, in terms of specific timing needs, addressed by specific levels of software (interrupt service routine versus user application level) . . .”

Read full story

 
This article was originally published on LinuxDevices.com and has been donated to the open source community by QuinStreet Inc. Please visit LinuxToday.com for up-to-date news and articles about Linux and open source.