Scribd
Upload
10 views

Scheduling Algorithms and Operating Systems

Uploaded by

Alvy Zaman
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
10 views

Scheduling Algorithms and Operating Systems

Uploaded by

Alvy Zaman
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Scheduling Algorithms and Operating Systems

Support for Real-Time Systems


by
KRITHI RAMAMRITHAM, MEMBER, IEEE, AND JOHN A. STANKOVIC, FELLOW, IEEE

Summary of the Research Article:

The comprehensive exploration of real-time operating systems (RTOS) in this text navigates
through challenges and progressions within the domain. The critique of RT-UNIX and RT-MACH
highlights hurdles in scaling proprietary kernels for larger applications and the tendency of time-
sharing extensions to prioritize speed over predictability. While these systems find applicability in
scenarios with less severe consequences for missed deadlines, guaranteeing performance becomes
intricate in intricate scenarios, demanding meticulous priority design and incurring substantial
costs.
The research landscape unfolds across three principal categories: small, fast, proprietary kernels;
real-time extensions to commercial operating systems; and research operating systems. Small
kernels, whether proprietary or commercially available, emerge as favored choices for small
embedded systems demanding swift, predictable execution. Nevertheless, challenges surface as
application complexity grows. Real-time extensions to commercial systems, exemplified by RT-
UNIX and RT-MACH, grapple with difficulties in transforming non-real-time systems, often
misaligning with real-time constraints.
Within the realm of research operating systems, there is a concerted effort to challenge
fundamental assumptions inherent in time-sharing systems. Ongoing trends encapsulate the
development of new paradigms, real-time process models, synchronization primitives, fault
tolerance mechanisms, and support for multiprocessor and distributed real-time systems. The
primary emphasis is on predictability, extending beyond the kernel to support application-level
predictability. Several noteworthy research projects showcase diverse approaches.

Categories of Research Operating Systems:


1. Small, Fast, Proprietary Kernels:
• Used for small embedded systems where fast and highly predictable execution is
crucial.
• Challenges in maintaining predictability as applications become more complex.
2. Real-Time Extensions to Commercial Operating Systems:
• Extensions like RT-UNIX, RT-POSIX, RT-MACH to commercial systems.
• Slower and less predictable than proprietary kernels but offer greater functionality
and better development environments.
• Challenges in converting non-real-time systems to real-time versions.
3. Research Operating Systems:
• Ongoing research focuses on challenging basic assumptions made by time-sharing
operating systems.
• Trends include new paradigms, real-time process models, synchronization
primitives, fault tolerance, object-oriented approaches, and support for
multiprocessor and distributed real-time systems.
• Emphasis on predictability of both the kernel and application-level predictability.

MARS (Microprocessor and Real-time Systems) introduces a paradigm shift by adopting a


time-driven model and static table-driven scheduling. The emphasis on automatic flow control and
robust support for distributed real-time systems distinguishes MARS as a pioneering endeavor.
The Spring kernel tackles the challenges of multiprocessors and distributed systems with its
dynamic planning-based scheduling and a reflective architecture, offering a high degree of
flexibility. MARUTI, in turn, places a spotlight on dynamic on-line guarantees and fault tolerance,
employing an object-based system with a top-down design philosophy.
ARTS (Ada Real-Time System) chart a course toward a distributed real-time computing
environment, leveraging real-time objects and threads. The integration of a priori schedulability
analysis marks ARTS as a noteworthy endeavor. CHAOS adopts an object-based approach,
supporting a versatile family of kernels and embracing a guarantee-oriented scheduling paradigm.
HARTOS (Hexagonal Architecture for Real-Time Systems) unfolds within a hexagonal mesh
network, addressing inter-process communication intricacies and furnishing support for distributed
real-time systems.

General Trends in Research:


• New approaches challenging time-sharing assumptions.
• Focus on predictability at both the kernel and application levels.
• Exploration of object-oriented approaches and support for fault tolerance.
• Interest in multiprocessor and distributed real-time systems.
In conclusion, the exploration of real-time operating systems (RTOS) in this discourse unveils a
multifaceted landscape replete with challenges, innovations, and ongoing endeavors. The critical
examination of RT-UNIX and RT-MACH underscores the nuanced intricacies faced in adapting
proprietary kernels to larger applications and the inherent trade-off between speed and
predictability in time-sharing extensions. While these systems find a niche in scenarios with lenient
repercussions for missed deadlines, ensuring robust performance becomes a labyrinthine task in
intricate environments, necessitating meticulous priority design and incurring substantial costs.
The research panorama unfolds across three cardinal categories, each delineating distinct
trajectories: small, fast, proprietary kernels; real-time extensions to commercial operating systems;
and the avant-garde research operating systems. Small kernels, whether of proprietary or
commercial origin, emerge as preferred choices for petite embedded systems demanding
expeditious, predictable execution. However, the crescendo of challenges becomes palpable as the
symphony of application complexity gains momentum. Real-time extensions to commercial
systems, exemplified by RT-UNIX and RT-MACH, grapple with the formidable task of
transforming non-real-time systems, often at odds with the stringent demands of real-time
constraints.

You might also like