Alternatives for scheduling virtual machines in real-time embedded systems
Proceedings of the 1st workshop on Isolation and integration in embedded systems
| Hi-index | 0.00 |
Today's technology is changing and improving at such a rapid rate that even the most state of the art systems seem to be obsolete before completion. Sometimes, changing customer requirements dictate that a smaller, faster, and more powerful system be produced to perform the same tasks. Target migration, the transfer of a software product from one target system to another more powerful target system, is one cost effective way to upgrade system performance.Target migration requires careful analysis to determine the impact of the new target environment on system performance. In many cases, the goal of target migration is to improve processor loading and throughput, while reducing memory requirements. A specific methodology needs to be established that will enable the design team to perform the required analysis and then be able to identify and measure the improvements on the system performance and memory usage. This methodology must be established early by the design team and then followed through all phases of the target migration effort.This paper describes methodology used by a Martin Marietta (formerly GE Aerospace) design team that was asked to determine the impact of a target migration to the MIL-STD-1750A Architecture. The methodology is divided into three phases. The steps to be performed in each phase are described and examples are provided from an actual project. The lessons learned from our experiences with this new methodology will benefit other design teams tasked with migrating existing software products to smaller but more powerful target systems.