Analytical Modeling for Operating System Schedulers on NUMA Systems

Chanin, Rafael; Corrêa, M.; Fernandes, Paulo; Sales, Afonso; Scheer, R.; Zorzo, Avelino F.

doi:10.1016/j.entcs.2006.03.016

Cited by 12 publications

(9 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…However, the results of the simulation and formal analysis are contradicting. While the simulation and measurements yielded a performance gain of 2.2% to 10% depending on the underlying hardware architecture [18], the analytical results predicted an improvement of at most 1% [17]. The authors do not address the differences in the results of the predictions and measurements in the mentioned papers.…”

Section: Related Workmentioning

confidence: 49%

“…General Purpose Operating Systems: Chanin, Correa et al [17], [18] analysed the influence of different load balancing polices for Non-Uniform Memory Access (NUMA) systems on software performance focussing on the effect of different memory access times. They proposed an optimised multilevel load balancing algorithm and demonstrated with simulations, measurements and formal analyses of stochastic automata networks the possible performance gain of the new algorithm.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Performance evaluation of scheduling policies in symmetric multiprocessing environments

Happe

Groenda

Reussner

2009

2009 IEEE International Symposium on Modeling, Analysis &Amp; Simulation of Computer and Telecommunication Systems

View full text Add to dashboard Cite

Abstract-The shift of hardware architecture towards parallel execution led to a broad usage of multi-core processors in desktop systems and in server systems. The benefit of additional processor cores for software performance depends on the software's parallelism as well as the operating system scheduler's capabilities. Especially, the load on the available processors (or cores) strongly influences response times and throughput of software applications. Hence, a sophisticated understanding of the mutual influence of software behaviour and operating system schedulers is essential for accurate performance evaluations. Multi-core systems pose new challenges for performance analysis and developers of operating systems. For example, an optimal scheduling policy for multi-server systems, such as shortest remaining processing time (SRPT) for single-server systems, is not yet known in queueing theory. In this paper, we present a detailed experimental evaluation of general purpose operating system (GPOS) schedulers in symmetric multiprocessing (SMP) environments. In particular, we are interested in the influence of multiprocessor load balancing on software performance. Additionally, the evaluation includes effects of GPOS schedulers that can also occur in single-processor environments, such as I/Oboundedness of tasks and different prioritisation strategies. The results presented in this paper provide the basis for the future development of more accurate performance models of today's software systems.

show abstract

Section: Related Workmentioning

confidence: 49%

Section: Related Workmentioning

confidence: 99%

Performance evaluation of scheduling policies in symmetric multiprocessing environments

Happe

Groenda

Reussner

2009

2009 IEEE International Symposium on Modeling, Analysis &Amp; Simulation of Computer and Telecommunication Systems

View full text Add to dashboard Cite

show abstract

“…The product state space is given by [3 (S+1) × (K + 1)]. (v) Non-Uniform Memory Access (NUMA) model [9] -a model of processes running in NUMA processors for the Linux operating system. NUMA is a model for capturing the behavior of processes and processors in the Operating System, to analytically calculate the chances for a given processor to fail.…”

Section: Numerical Analysismentioning

confidence: 99%

Efficient vector-descriptor product exploiting time-memory trade-offs

Czekster

Fernandes

Webber

2011

SIGMETRICS Perform. Eval. Rev.

Self Cite

View full text Add to dashboard Cite

The description of large state spaces through stochastic structured modeling formalisms like stochastic Petri nets, stochastic automata networks and performance evaluation process algebra usually represent the infinitesimal generator of the underlying Markov chain as a Kronecker descriptor instead of a single large sparse matrix. The best known algorithms used to compute iterative solutions of such structured models are: the pure sparse solution approach, an algorithm that can be very time efficient, and almost always memory prohibitive; the Shuffle algorithm which performs the product of a descriptor by a probability vector with a very impressive memory efficiency; and a newer option that offers a trade-off between time and memory savings, the Split algorithm. This paper presents a comparison of these algorithms solving some examples of structured Kronecker represented models in order to numerically illustrate the gains achieved considering each model's characteristics.

show abstract

“…The current O(1) scheduler keeps a runqueue per processor, which is responsible for containing all the executable processes of a given processor. Thus, if a process is inserted in a runqueue of a specific processor, it will only run on that processor [6]. Each runqueue contains two priority arrays [7]: active and expired.…”

Section: Linux Schedulermentioning

confidence: 99%

“…However, very few discuss the behavior of the front-end computers when the traffic turns into processes in a given operation system. For example, [6] presents a generic model for an Operating System (OS) scheduler for Non-Uniform Memory Access (NUMA) machines using the Stochastic Automata Networks (SAN) formalism. SAN is used to describe processes and processors in the OS and their behavior when processes have to be migrated.…”

Section: Introductionmentioning

confidence: 99%

A Markovian Performance Model for Resource Allocation Scheduling on GNU/Linux

Kawasaki

Guedes

Cardoso

et al. 2006

Frontiers of High Performance Computing and Networking – ISPA 2006 Workshops

View full text Add to dashboard Cite

Abstract. The current paper addresses the problem of quality of service (QoS) provisioning in a general purpose operating system (GPOS), in this case GNU/Linux. Particularly, we propose to change the CPU allocation in that OS by reserving a percentage of a CPU capacity in order to ensure the QoS provisioning according to the QoS demand of each process. In order to investigate the effectiveness of that approach, Markovian models are proposed to represent the dynamics of the systems. Results show that the OS with reservation outperforms the system without it, but also that there is a performance tradeoff in the OS with reservation in such a way that an improvement in the QoS perceived by processes using the reserved capacity is done at a cost of a degradation in the QoS perceived by the other processes.

show abstract

Analytical Modeling for Operating System Schedulers on NUMA Systems

Cited by 12 publications

References 10 publications

Performance evaluation of scheduling policies in symmetric multiprocessing environments

Performance evaluation of scheduling policies in symmetric multiprocessing environments

Efficient vector-descriptor product exploiting time-memory trade-offs

A Markovian Performance Model for Resource Allocation Scheduling on GNU/Linux

Contact Info

Product

Resources

About