Performance Evaluation of the XDEM framework on the OpenStack Cloud - - PowerPoint PPT Presentation

▶
performance evaluation of the xdem framework on the
SMART_READER_LITE
LIVE PREVIEW

Performance Evaluation of the XDEM framework on the OpenStack Cloud - - PowerPoint PPT Presentation

Dec 07, 2023 268 likes •475 views

Introduction & Context Performance Evaluation Conclusions Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware X. Besseron 1 V. Plugaru 2 A. H. Mahmoudi 1 S. Varrette 2 B. Peters 1 . Bouvry 2 P 1

slide-1
SLIDE 1

Introduction & Context Performance Evaluation Conclusions

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware

  • X. Besseron1
  • V. Plugaru2
  • A. H. Mahmoudi1
  • S. Varrette2
  • B. Peters1

P . Bouvry2

1Luxembourg XDEM Research Centre (LuXDEM) 2Parallel Computing and Optimisation Group (PCOG)

Faculty of Science, Technology and Communication University of Luxembourg, Luxembourg

SC-Camp 2015

(initially presented at PARENG 2015)

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 1 / 17

slide-2
SLIDE 2

Introduction & Context Performance Evaluation Conclusions

Outline

1

Introduction & Context eXtended Discrete Element Method Cloud Computing

2

Performance Evaluation Methodology Experimental Results

3

Conclusions

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 2 / 17

slide-3
SLIDE 3

Introduction & Context Performance Evaluation Conclusions

Cloud Computing for Discrete Element Method?

Processing of granular materials

  • Snow, sand, gravel, coke, iron oxide,

biomass, food, tablets, ...

  • Widely used in industry
  • Discrete Element Method (DEM)

DEM, HPC and Cloud Computing

  • Huge computation time ⇒ Parallel execution required
  • Traditionally addressed using High Performance Computing

(HPC) platforms

֒ → Cloud Computing (CC) appears as a promising alternative

= ⇒ How suitable is the Cloud Computing approach for an HPC workflow such as a DEM application?

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 3 / 17

slide-4
SLIDE 4

Introduction & Context Performance Evaluation Conclusions

Cloud Computing for Discrete Element Method?

Processing of granular materials

  • Snow, sand, gravel, coke, iron oxide,

biomass, food, tablets, ...

  • Widely used in industry
  • Discrete Element Method (DEM)

DEM, HPC and Cloud Computing

  • Huge computation time ⇒ Parallel execution required
  • Traditionally addressed using High Performance Computing

(HPC) platforms

֒ → Cloud Computing (CC) appears as a promising alternative

= ⇒ How suitable is the Cloud Computing approach for an HPC workflow such as a DEM application?

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 3 / 17

slide-5
SLIDE 5

Introduction & Context Performance Evaluation Conclusions

eXtended Discrete Element Method (XDEM)

XDEM software

  • numerical simulation framework
  • extends the classic DEM

approach

  • parallel execution with MPI

Multi-physics simulation

  • Particle motion
  • Chemical conversion
  • Finite Element Method (FEM)

coupling (with Diffpack)

  • Computational Fluid Dynamics

(CFD) coupling (with OpenFoam)

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 4 / 17

slide-6
SLIDE 6

Introduction & Context Performance Evaluation Conclusions

XDEM example: Blast Furnace

Used in metallurgy to produce hard metals

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 5 / 17

slide-7
SLIDE 7

Introduction & Context Performance Evaluation Conclusions

Cloud Computing

On-demand, online access to computing resources and services

What kind of services?

  • Software, Platform as a Service (SaaS, PaaS)
  • Infrastructure as a Service (IaaS)

֒ → i.e. deploy your own OS, software layer and applications

IaaS relies on a virtualization layer

  • Provisions user virtual machines on-demand
  • Provides flexibility yet adds overhead to operations
  • Hypervisors: Xen, KVM, VMWare ESXi, Microsoft Hyper-V, ...
  • Cloud middleware: OpenStack, Eucalyptus, Nimbus,

OpenNebula, VMWare vCloud, ...

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 6 / 17

slide-8
SLIDE 8

Introduction & Context Performance Evaluation Conclusions

Current study

Objectives

  • Study the overhead of Cloud Computing middleware on an

HPC workflow

  • Extend previous work [1] to a real application

Systematic Performance Evaluation

  • Fair comparison using the exact same hardware
  • Large scale distributed execution totalling hundreds of cores
  • Real application with XDEM and a real-life test case
  • Automated experimental framework and reproducible measurements

[1]

  • S. Varrette, V. Plugaru, M. Guzek, X. Besseron, P

. Bouvry HPC Performance and Energy-Efficiency of the OpenStack Cloud Middleware Heterogeneous and Unconventional Cluster Architectures and Applications Workshop (HUCAA’14)

  • Proc. of the 43rd Intl. Conf. on Parallel Processing (ICPP-2014)

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 7 / 17

slide-9
SLIDE 9

Introduction & Context Performance Evaluation Conclusions

Grid5000 & Kadeploy

Grid’5000

  • Large scale nation wide infrastructure

֒ → 8 sites in France, 1 in Luxembourg

  • 23 clusters, 941 nodes, 7494 cores
  • Designed for large scale parallel and

distributed computing research

Kadeploy

  • Scalable, efficient and reliable deployment system
  • Used as bare-metal provisioning solution
  • Many OS environments pre-defined, easily customizable
  • Integrates with KaVLAN to deploy inside isolated, routed or

grid-global VLANs

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 8 / 17

slide-10
SLIDE 10

Introduction & Context Performance Evaluation Conclusions

Experimental setup

Three configurations

1 Native (no virtualization) 2 OpenStack with Xen hypervisor 3 OpenStack with KVM hypervisor

Two clusters

  • PetitPrince: Intel-based, 10-Gigabit Ethernet
  • StRemi: AMD-based, 1-Gigabit Ethernet

Computing and networking performance

  • Performance metric: XDEM iteration time

֒ → Reported value: average of at least 20 measurements

  • Only one virtual machine per physical node
  • In-memory file reading/writing operations

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 9 / 17

slide-11
SLIDE 11

Introduction & Context Performance Evaluation Conclusions

Test case: Biomass pyrolysis with XDEM

Wood decomposition reactions

Wood → Char Wood → Tar Wood → νCO CO + νCO2 CO2 + νH2O H2O + νH2 H2 + νCH4 CH4

Initial conditions

  • Wood packed bed of 19 cm
  • 32.000 spherical particles

with diameter of 6.2 mm

  • Wood particles initially at

363 K with 8% wb moisture

  • Surrounding gas temperature of

1200 K and pressure of 1 bar

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 10 / 17

slide-12
SLIDE 12

Introduction & Context Performance Evaluation Conclusions

Single-core performance

  • Sequential execution, only one process with one thread
  • Overhead of virtualization on computing performance

3.45 s 3.54 s 3.80 s + 2.8 % + 10.4 %

1 2 3 4 Native OpenStack/KVM OpenStack/XEN

Average Iteration Time (s) (Lower is better)

PetitPrince cluster 6.54 s 7.22 s 6.78 s + 10.4 % + 3.7 %

2 4 6 Native OpenStack/KVM OpenStack/XEN

Average Iteration Time (s) (Lower is better)

StRemi cluster

= ⇒ Overhead between 2.8 % and 10.4 %

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 11 / 17

slide-13
SLIDE 13

Introduction & Context Performance Evaluation Conclusions

Single-node performance

  • Execution on one full node, one process per core
  • MPI communications using shared memory

1 2 3 1 2 4 8 12

Number of Processes (1 node) Average Iteration Time (s) (Lower is better)

Native OpenStack/KVM OpenStack/XEN

PetitPrince cluster

2 4 6 1 2 4 8 12 16 20 24

Number of Processes (1 node) Average Iteration Time (s) (Lower is better)

Native OpenStack/KVM OpenStack/XEN

StRemi cluster

KVM overhead XEN overhead PetitPrince (12 processes) 3.5% 6.4% StRemi (24 processes) 14.0% 8.1%

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 12 / 17

slide-14
SLIDE 14

Introduction & Context Performance Evaluation Conclusions

Internode communication

  • OSU Micro-Benchmarks
  • MPI internode bandwidth
  • Two processes on two different nodes
  • 300

600 900 1200 10 1,000 100,000

Message Size (Bytes) − LOGSCALE Bandwidth (MBytes/s) (Higher is better)

  • Native

OpenStack/KVM OpenStack/XEN

PetitPrince cluster

  • 30

60 90 120 10 1,000 100,000

Message Size (Bytes) − LOGSCALE Bandwidth (MBytes/s) (Higher is better)

  • Native

OpenStack/KVM OpenStack/XEN

StRemi cluster

  • Virtualized environments cannot sustain more than 25 % of

the available bandwidth

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 13 / 17

slide-15
SLIDE 15

Introduction & Context Performance Evaluation Conclusions

Multi-node performance

  • Distributed execution, one process per core
  • MPI communication using Ethernet network between nodes
  • Speedup = T Native

seq

/Tp

  • 3

6 9 25 50 75 100 125

Number of Processes (12 processes/node) Speedup Tseq Tp (Higher is better)

  • Native

OpenStack/KVM OpenStack/XEN

PetitPrince cluster

  • 4

8 12 16 100 200 300 400

Number of Processes (24 processes/node) Speedup Tseq Tp (Higher is better)

  • Native

OpenStack/KVM OpenStack/XEN

StRemi cluster

= ⇒ OpenStack/KVM performs better than OpenStack/Xen

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 14 / 17

slide-16
SLIDE 16

Introduction & Context Performance Evaluation Conclusions

Multi-node performance

Summary for the best configuration in the scalability range

PetitPrince StRemi 10 nodes, 120 processes 8 nodes, 192 processes Iteration Time Speedup Iteration Time Speedup Native 0.32 s 10.8 0.41 s 15.9 OpenStack/KVM 0.36 s + 12.6% 9.6 0.53 s + 28.5% 12.4 OpenStack/XEN 0.51 s + 60.1% 6.7 0.55 s + 33.4% 11.9

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 15 / 17

slide-17
SLIDE 17

Introduction & Context Performance Evaluation Conclusions

Conclusions & Future Works

Conclusions – HPC vs CC

  • Sequential executions: moderate overhead < 11 %
  • Distributed executions: moderate (10 %) to significant

(60 %) overhead

  • Communication appears to be the bottleneck

֒ → study limited to Ethernet (no InfiniBand)

  • Cloud Computing offers other advantages

֒ → on-demand provisioning, cost reduction, ...

Future Works

  • Other test cases, other applications
  • Newer versions of OpenStack, Xen, KVM
  • Single Root I/O Virtualization (SR-IOV) technology

(promising for InfiniBand)

  • LinuX Containers (LXC)

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 16 / 17

slide-18
SLIDE 18

Introduction & Context Performance Evaluation Conclusions

Thank you for your attention!

Questions?

Luxembourg XDEM Research Centre (LuXDEM) http://luxdem.uni.lu Parallel Computing and Optimisation Group (PCOG) http://pcog.uni.lu

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 17 / 17

slide-19
SLIDE 19

Intranode communication

  • OSU Micro-Benchmarks
  • MPI intranode bandwidth
  • Two processes on the same node
  • 1000

2000 10 1,000 100,000

Message Size (Bytes) − LOGSCALE Bandwidth (MBytes/s) (Higher is better)

  • Native

OpenStack/KVM OpenStack/XEN

PetitPrince cluster

  • 250

500 750 1000 1250 10 1,000 100,000

Message Size (Bytes) − LOGSCALE Bandwidth (MBytes/s) (Higher is better)

  • Native

OpenStack/KVM OpenStack/XEN

StRemi cluster

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 1 / 2

slide-20
SLIDE 20

Hardware and software configuration

Cluster PetitPrince StRemi Site Luxembourg Reims Processor type Intel Xeon AMD Opteron Processor model E5-2630L @ 2GHz 6164 HE @ 1.7GHz #nodes 16 44 #CPUs per node 2 2 #cores per node 6 12 Memory per node 32 GBytes 48 GBytes Network 10-Gigabit Ethernet 1-Gigabit Ethernet Operating System (Hyp.) Ubuntu 12.04 LTS, Linux 3.2 Operating System (VM) Debian 7.1, Linux 3.2 Cloud Middleware OpenStack Essex OpenMPI 1.4.3 OSU Micro Benchmark 4.4.1 XDEM software Internal v2015.01.05

Performance Evaluation of the XDEM framework on the OpenStack Cloud Computing Middleware 2 / 2