Real-Time Systems has announced a new version of their Hypervisor, now allowing users to prioritize simultaneous access of guest operating systems to shared resources like shared caches or the main memory itself. This new approach avoids lower priority operating systems influencing an RTOS using the same the last level cache.
Strict separation of operating systems while preserving hard real-time determinism has been one of the strongest attributes of the RTS Hypervisor all along. While so far virtualization technology has been used to virtualize and monitor unmodified, non-real-time operating systems like Microsoft Windows, RTS now takes it one step further. As an industry first, the RTS Hypervisor now detects excessive memory usage of a non-real-time Operating System, which due to shared utilization of cache or simultaneous access to the system's main memory could cause jitter in a high priority real-time operating system running in parallel.
A finely adjustable throttle can now limit the amount of memory throughput available to a virtualized operating system in order to avoid negative impact on an RTOS due to potential cache misses or an occupied data bus. This throttle only takes effect if a user configurable threshold is about to be reached. Relevant tests show that this new feature greatly reduces jitter and therefore improves determinism significantly. Now customers with very demanding requirements regarding latencies and determinism can finally also run multiple operating systems on the same hardware while still providing the hard real-time capability expected of their system.
The RTS Hypervisor allows customers to partition an Intel x86 multi core platform into virtually independent computers running e.g. a Microsoft operating system as a user interface and separate real-time operating systems in parallel for real-time tasks on the same hardware. Consolidating hardware greatly reduces cost while increasing reliability and reducing power consumption. The hypervisor solution is particularly useful for products in the industrial, medical or transportation markets where many customers rely on a Real-Time Operating System (RTOS) for time or mission critical functions while using a different operating system (OS) for other tasks, such as running a human machine interface (HMI). Because the RTS solution does not depend on a host OS, operating systems can start in any sequence, reboot independently and never harm each other.
Real-Time Systems at Embedded World 2014: halle 5 / stand 468