A Three-Level Modelling Approach for Asynchronous Speed Scaling in High-Performance Data Centres

In data centres, there exist several techniques for energy efficiency purposes. When applied, most of those techniques have impact on the quality (e.g. performance) of the underlying services. A careful study is required in order to optimise such an energy-performance trade-off. In this paper, we study the speed scaling as an energy efficiency technique within the scope of high-performance computing (HPC) data centres. We propose a methodology based on three-levels: analytic, simulation and technical. At the analytical level, the matrix-analytic method (MAM) allows one to obtain energy-performance measures explicitly for a small-scale system. At the simulation level, discrete-event simulation (DES) based on the generalised semi-Markov processes (GSMP) is used to derive the corresponding estimates. Finally, at the technical level, a real small-scaled system in a controlled environment is used. The preliminary results demonstrate that simulation and technical models go well together with the theoretical one with an accuracy of more than 95%.

[1]  Alexander S. Rumyantsev,et al.  Stability criterion of a multiserver model with simultaneous service , 2017, Ann. Oper. Res..

[2]  P. Glynn.  A GSMP formalism for discrete event systems , 1989, Proc. IEEE.

[3]  Asadullah Shah,et al.  A comparative analysis of resource allocation schemes for real-time services in high-performance computing systems , 2020, Int. J. Distributed Sens. Networks.

[4]  Abdullah Muhammed,et al.  An Adaptive Energy Allocation for High-Performance Computing Systems Using a Cyber-Physical Approach , 2017 .

[5]  Mor Harchol-Balter,et al.  Open problems in queueing theory inspired by datacenter computing , 2021, Queueing Syst. Theory Appl..

[6]  Prashant Pandey,et al.  Cloud computing , 2010, ICWET.

[7]  Mikko Majanen,et al.  Energy-aware job scheduler for high-performance computing , 2012, Computer Science - Research and Development.

[8]  Hermann de Meer,et al.  Modelling and Analysing Conservative Governor of DVFS-enabled Processors , 2018, e-Energy.

[9]  Enrico Calore,et al.  Performance and Power Analysis of HPC Workloads on Heterogenous Multi-Node Clusters , 2018 .

[10]  Helen D. Karatza,et al.  An M/M/2 parallel system model with pure space sharing among rigid jobs , 2007, Math. Comput. Model..

[11]  Peter J. Haas,et al.  On Transience and Recurrence in Irreducible Finite-State Stochastic Systems , 2015, ACM Trans. Model. Comput. Simul..

[12]  Robert Basmadjian,et al.  Flexibility-Based Energy and Demand Management in Data Centers: A Case Study for Cloud Computing , 2019, Energies.

[13]  Qi-Ming He,et al.  Fundamentals of Matrix-Analytic Methods , 2013, Springer New York.

[14]  Marcel F. Neuts,et al.  Matrix-geometric solutions in stochastic models - an algorithmic approach , 1982 .

[15]  Paul J. Kühn,et al.  DVFS-Power Management and Performance Engineering of Data Center Server Clusters , 2019, 2019 15th Annual Conference on Wireless On-demand Network Systems and Services (WONS).

[16]  Hermann de Meer,et al.  A Generic Methodology to Derive Empirical Power Consumption Prediction Models for Multi-Core Processors , 2013, 2013 International Conference on Cloud and Green Computing.

[17]  Xue Liu,et al.  Dynamic Voltage Scaling in Multitier Web Servers with End-to-End Delay Control , 2007, IEEE Transactions on Computers.

[18]  Tom Burr,et al.  Introduction to Matrix Analytic Methods in Stochastic Modeling , 2001, Technometrics.

[19]  Kevin Skadron,et al.  Multi-mode energy management for multi-tier server clusters , 2008, 2008 International Conference on Parallel Architectures and Compilation Techniques (PACT).

[20]  Alexander S. Rumyantsev,et al.  On an exact solution of the rate matrix of G∕M∕1-type Markov process with small number of phases , 2018, J. Parallel Distributed Comput..

[21]  Xue Liu,et al.  OptiTuner: On Performance Composition and Server Farm Energy Minimization Application , 2011, IEEE Transactions on Parallel and Distributed Systems.

[22]  Marcel F. Neuts,et al.  Matrix-Geometric Solutions in Stochastic Models , 1981 .

[23]  Alexander S. Rumyantsev,et al.  Evaluating a Single-Server Queue with Asynchronous Speed Scaling , 2018, MMB.

[24]  A. Rumyantsev.  Stability of Multiclass Multiserver Models with Automata-type Phase Transitions , 2020 .

[25]  Peter G. Harrison,et al.  Energy--Performance Trade-Offs via the EP Queue , 2016, ACM Trans. Model. Perform. Evaluation Comput. Syst..

[26]  Dzmitry Kliazovich,et al.  Green Data Centers , 2015 .

[27]  Hermann de Meer,et al.  Modelling performance and power consumption of utilisation-based DVFS using M/M/1 queues , 2016, e-Energy.

[28]  Marie Frei,et al.  Analyzing Markov Chains Using Kronecker Products Theory And Applications , 2016 .

[29]  Anders S. G. Andrae,et al.  On Global Electricity Usage of Communication Technology: Trends to 2030 , 2015 .

[30]  BasmadjianRobert,et al.  Energy-aware job scheduler for high-performance computing , 2012 .