Cost-Benefit Analysis of Cloud Computing versus Desktop Grids



Similar documents
Volunteer Computing, Grid Computing and Cloud Computing: Opportunities for Synergy. Derrick Kondo INRIA, France

Volunteer Computing and Cloud Computing: Opportunities for Synergy

Cost-Benefit Analysis of Cloud Computing versus Desktop Grids. By : Paritosh Heera( MT )

How To Compare Amazon Ec2 To A Supercomputer For Scientific Applications

Part V Applications. What is cloud computing? SaaS has been around for awhile. Cloud Computing: General concepts

XMPP A Perfect Protocol for the New Era of Volunteer Cloud Computing

System Models for Distributed and Cloud Computing

Manjrasoft Market Oriented Cloud Computing Platform

Concepts Introduced in Chapter 6. Warehouse-Scale Computers. Important Design Factors for WSCs. Programming Models for WSCs

Using Proxies to Accelerate Cloud Applications

Manjrasoft Market Oriented Cloud Computing Platform

Building a Volunteer Cloud

CHAPTER 8 CLOUD COMPUTING

wu.cloud: Insights Gained from Operating a Private Cloud System

What Is It? Business Architecture Research Challenges Bibliography. Cloud Computing. Research Challenges Overview. Carlos Eduardo Moreira dos Santos

Practical Approach for Achieving Minimum Data Sets Storage Cost In Cloud

Elastic Cloud Computing in the Open Cirrus Testbed implemented via Eucalyptus

How Liferay Is Improving Quality Using Hundreds of Jenkins Servers

Introduction to Cloud Computing

Cloud Computing with Red Hat Solutions. Sivaram Shunmugam Red Hat Asia Pacific Pte Ltd.

The Cisco Powered Network Cloud: An Exciting Managed Services Opportunity

Navigating Among the Clouds. Evaluating Public, Private and Hybrid Cloud Computing Approaches

Grid Computing vs Cloud

Graph Analytics in Big Data. John Feo Pacific Northwest National Laboratory

The Hidden Extras. The Pricing Scheme of Cloud Computing. Stephane Rufer

CLUSTER COMPUTING TODAY

Parametric Analysis of Mobile Cloud Computing using Simulation Modeling

Chapter 7: Distributed Systems: Warehouse-Scale Computing. Fall 2011 Jussi Kangasharju

Cloud Computing with Microsoft Azure

Building a Private Cloud with Eucalyptus

What is Cloud Computing? Tackling the Challenges of Big Data. Tackling The Challenges of Big Data. Matei Zaharia. Matei Zaharia. Big Data Collection

Cloud Computing. Adam Barker

Introduction to Big Data! with Apache Spark" UC#BERKELEY#

Cloud Computing- Research Issues and Challenges

Maginatics Cloud Storage Platform for Elastic NAS Workloads

Chapter 4 Cloud Computing Applications and Paradigms. Cloud Computing: Theory and Practice. 1

Distributed Systems LEEC (2005/06 2º Sem.)

Cloud Computing Is In Your Future

DISTRIBUTED SYSTEMS AND CLOUD COMPUTING. A Comparative Study

Cloud Computing for Increased Business Value

A New Approach of CLOUD: Computing Infrastructure on Demand

The Lattice Project: A Multi-Model Grid Computing System. Center for Bioinformatics and Computational Biology University of Maryland

High Performance Computing Cloud Computing. Dr. Rami YARED

Sriram Krishnan, Ph.D.

Intel Platform and Big Data: Making big data work for you.

Chapter 19 Cloud Computing for Multimedia Services

CUMULUX WHICH CLOUD PLATFORM IS RIGHT FOR YOU? COMPARING CLOUD PLATFORMS. Review Business and Technology Series

Amazon EC2 Product Details Page 1 of 5

GIS IN THE CLOUD THE ESRI EXAMPLE DAVID CHAPPELL SEPTEMBER 2010 SPONSORED BY ESRI

WORKFLOW ENGINE FOR CLOUDS

A Service for Data-Intensive Computations on Virtual Clusters

Cloud Computing Submitted By : Fahim Ilyas ( ) Submitted To : Martin Johnson Submitted On: 31 st May, 2009

SURFsara HPC Cloud Workshop

Amazon Web Services 100 Success Secrets

Security Benefits of Cloud Computing

How To Evaluate The Cost Performance Of A Cloud Cache On A Microsoft Cloud Instance (Aws Cloud)

How To Understand Cloud Computing

Cloud Computing. Summary

Cloud Computing and Amazon Web Services

Grid Computing Perspectives for IBM

ediscovery and Search of Enterprise Data in the Cloud

HYPER-CONVERGED INFRASTRUCTURE STRATEGIES

CSE 590: Special Topics Course ( Supercomputing ) Lecture 10 ( MapReduce& Hadoop)

Large scale processing using Hadoop. Ján Vaňo

Oracle Applications and Cloud Computing - Future Direction

Towards Predictable Datacenter Networks

Network & HEP Computing in China. Gongxing SUN CJK Workshop & CFI

How In-Memory Data Grids Can Analyze Fast-Changing Data in Real Time

Putchong Uthayopas, Kasetsart University

IMCM: A Flexible Fine-Grained Adaptive Framework for Parallel Mobile Hybrid Cloud Applications

Cloud Design and Implementation. Cheng Li MPI-SWS Nov 9 th, 2010

Transcription:

Cost-Benefit Analysis of Cloud Computing versus Desktop Grids Derrick Kondo, Bahman Javadi, Paul Malécot, Franck Cappello INRIA, France David P. Anderson UC Berkeley, USA

Cloud Background Vision Hide complexity of hardware and software management from a user by offering computing as a service Benefits Pay as you go Scale up or down dynamically No hardware management, less software management

Platform Performance vs. Costs high Cost low low Performance high

Platform Performance vs. Costs high Supercomputer Grid Cluster Cost low low Performance high

Platform Performance vs. Costs high Supercomputer Grid Cluster Cost low low Performance Desktop Grids (DG) high

Platform Performance vs. Costs high Supercomputer Grid Cluster Cost low low Cloud Performance What is the relationship? Desktop Grids (DG) high

Platform Performance vs. Costs high Supercomputer Grid Cluster Cost low low Cloud Performance What is the relationship? Desktop Grids (DG) DG + Cloud? high

Outline Performance tradeoffs Monetary tradeoffs Client hosting Server hosting

Apples to Apples Loosely-coupled, high-throughput, compute-intensive applications Tightly-coupled, data-intensive real-time applications low complexity high complexity

Apples to Apples Loosely-coupled, high-throughput, compute-intensive applications Tightly-coupled, data-intensive real-time applications low complexity high complexity DG s

Apples to Apples Loosely-coupled, high-throughput, compute-intensive applications Tightly-coupled, data-intensive real-time applications low complexity high complexity DG s Clouds

Apples to Apples Loosely-coupled, high-throughput, compute-intensive applications Tightly-coupled, data-intensive real-time applications low complexity high complexity DG s Clouds Comparison assuming embarrassingly parallel, compute-intensive applications

Method Use real performance measurements Exported BOINC (middleware for desktop grids) project data Use real costs Large/small BOINC projects (SETI@home / XtremLab) Amazon Elastic Computing Cloud (EC2)

Platform Construction Stages of Project &

Stages of Project & Platform Construction Deployment

Stages of Project & Platform Construction Deployment Execution

Stages of Project & Platform Construction Deployment Execution Completion

Platform Construction Deployment Execution Completion How long before I get X TeraFLOPS?

Platform Construction Deployment Execution Completion How long before I get X TeraFLOPS? 10 6 10 3 Number of cloud nodes 10 5 10 4 10 2 10 1 TeraFLOPS 10 3 0 5 10 15 20 25 30 Months for registration

Platform Construction Deployment Execution Completion How long before I get X TeraFLOPS? Number of cloud nodes 10 6 10 5 10 4 Can get over 20 TeraFLOPS within 6 months 10 3 10 2 10 1 TeraFLOPS 10 3 0 5 10 15 20 25 30 Months for registration

Platform Construction Deployment Execution Completion How long before I get X TeraFLOPS? Strategy: Add to BOINC project list Press releases Forum Announcements Google Ad Sense Respond to users (leverage volunteers) Number of cloud nodes 10 6 10 5 10 4 Can get over 20 TeraFLOPS within 6 months 10 3 0 5 10 15 20 25 30 Months for registration 10 3 10 2 10 1 TeraFLOPS

Platform Construction Deployment Execution Completion How long to deploy my batch of tasks needing faster response time?

Platform Construction Deployment Execution Completion How long to deploy my batch of tasks needing faster response time? 7,82/9*,34.4:*,.;*:3)4,6<!" '!" &!"!!" ".!"!!.!"".4=6>6!""".4=6>6!"""".4=6>6!" #!" $!" % ()*+,-./0.1/2)34,,-.3/5,6

Platform Construction Deployment Execution Completion How long to deploy my batch of tasks needing faster response time? 7,82/9*,34.4:*,.;*:3)4,6<!" '!" &!"!!" "!"!!.!"".4=6>6!""".4=6>6!"""".4=6>6 For 1000 tasks, ~10 minutes with 10 5 hosts!" #!" $!" % ()*+,-./0.1/2)34,,-.3/5,6.

Platform Construction Deployment Execution Completion How long to deploy my batch of tasks needing faster response time? Strategy: Specify lower latency bounds [Heien et al.] 7,82/9*,34.4:*,.;*:3)4,6<!" '!" &!"!!" "!"!!.!"".4=6>6!""".4=6>6!"""".4=6>6 For 1000 tasks, ~10 minutes with 10 5 hosts!" #!" $!" % ()*+,-./0.1/2)34,,-.3/5,6.

Platform Construction Deployment Execution Completion

Platform Construction Deployment Execution Completion How many volunteer nodes are equivalent to 1 cloud node?

Platform Construction Deployment Execution Completion How many volunteer nodes are equivalent to 1 cloud node?

Platform Construction Deployment Execution Completion How many volunteer nodes are equivalent to 1 cloud node? 2.8 active volunteer hosts per 1 cloud node. (Total performance still orders of magnitude better)

Platform Construction Deployment Execution Completion How many volunteer nodes are equivalent to 1 cloud node? 2.8 active volunteer hosts per 1 cloud node. (Total performance still orders of magnitude better) Strategy: Use statistical prediction of availability

Platform Construction Deployment Execution Completion

Platform Construction Deployment Execution Completion How long should I wait for task completion?

Platform Construction Deployment Execution Completion How long should I wait for task completion?

Platform Construction Deployment Execution Completion How long should I wait for task completion? Median project latency bound: 9 days for 3.7 hour work unit (on 3GHz host). Ratio of lat. bound / exec time > 5. Good success rates: 96.1% of WCG tasks met out of 227,000 tasks

Platform Construction Deployment Execution Completion How long should I wait for task completion? Strategy: See BOINC Catalog for typical deadlines and compute/comm/mem ratios. Median project latency bound: 9 days for 3.7 hour work unit (on 3GHz host). Ratio of lat. bound / exec time > 5. Good success rates: 96.1% of WCG tasks met out of 227,000 tasks

Monetary Tradeoffs Client hosting on cloud Not worth it and never will Server hosting on the cloud Possible solution

Monthly Project Costs

EC2 Pricing

What about Client Hosting on the Cloud?

Cost of Clouds versus DG How long until DG s more cost effective than Clouds?

Cost of Clouds versus DG How long until DG s more cost effective than Clouds? 7,8./9,)/:;<( #!!## #!!#$ #!!#% #!!#' F#'!'/G*HI8.,,)/8*3,5 2'J&/-H*I3/8*3,56 '"&$/G*HI8.,,)/8*3,5 2#&#$/-H*I3/8*3,56 =.),>;0? @ABCDE*>, A7$ / #!!#" #!!#& /! " #! #" $! $" %! ()*+,-./01,/230456

Cost of Clouds versus DG How long until DG s more cost effective than Clouds? 7,8./9,)/:;<( #!!## #!!#$ #!!#% #!!#' F#'!'/G*HI8.,,)/8*3,5 2'J&/-H*I3/8*3,56 '"&$/G*HI8.,,)/8*3,5 2#&#$/-H*I3/8*3,56 =.),>;0? @ABCDE*>, A7$ / #!!#" After ~13 days #!!#& /! " #! #" $! $" %! ()*+,-./01,/230456

Cost of Clouds versus DG (2) What are total costs over time?

Cost of Clouds versus DG (2) 8 x 104 What are total costs over time? Total cost 7 6 5 4 3 2 1 EC2 100n EC2 1000n EC2 10000n XtremLab SETI@home 0 0 5 10 15 20 Platform age (days)

Cost of Clouds versus DG (2) 8 x 104 What are total costs over time? Total cost 7 6 5 4 3 2 1 After 3 days, DG is cheaper EC2 100n EC2 1000n EC2 10000n XtremLab SETI@home 0 0 5 10 15 20 Platform age (days)

Cost of Clouds versus DG (3) How many months of DG can X months of Cloud buy me?

Cost of Clouds versus DG (3) How many months of DG can X months of Cloud buy me? Months with VC 3500 3000 2500 2000 1500 1000 Xtremlab / 1000n EC2 SETI@home / 1000n EC2 Xtremlab / 10000n EC2 SETI@home / 10000n EC2 500 0 0 5 10 15 20 25 Months with EC2

Cost of Clouds versus DG (3) How many months of DG can X months of Cloud buy me? Months with VC 3500 3000 2500 2000 1500 1000 Xtremlab / 1000n EC2 SETI@home / 1000n EC2 Xtremlab / 10000n EC2 SETI@home / 10000n EC2 12 months of Cloud can pay for > 125 DG years. DG way more sustainable. 500 0 0 5 10 15 20 25 Months with EC2

Equivalent Clouds given BOINC Costs Given BOINC costs, what size Cloud can I buy?

Equivalent Clouds given BOINC Costs Given BOINC costs, what size Cloud can I buy?

Equivalent Clouds given BOINC Costs 1.75 TeraFLOPS, 7 TeraBytes Given BOINC costs, what size Cloud can I buy?.75 TeraFLOPS, 3 TeraBytes

Equivalent Clouds given BOINC Costs Given BOINC costs, what size Cloud can I buy? 1.75 TeraFLOPS, 7 TeraBytes 2 orders of magnitude lower than BOINC.75 TeraFLOPS, 3 TeraBytes

What about Server Hosting on the Cloud?

Number of Hosts over Time

Number of Hosts over Time Load variation exits w/ publicity, projects run out of work, etc. Clouds take care of server management. Potential to exploit clouds.

Server Costs on a Cloud How much to host BOINC server on cloud?

Server Costs on a Cloud How much to host BOINC server on cloud?

Server Costs on a Cloud How much to host BOINC server on cloud? (versus 43K +12K/month) (versus {4,1}K + {5,1}K/month)

Server Costs on a Cloud How much to host BOINC server on cloud? 1.7-16.5 times cheaper to host on Cloud, but bandwidth is expensive. (versus 43K +12K/month) (versus {4,1}K + {5,1}K/month)

Cloud Resources with Given Budget How big of a server can I support with given budget?

Cloud Resources with Given Budget How big of a server can I support with given budget?

Cloud Resources with Given Budget How big of a server can I support with given budget? 4 Mbps, 2 TB 12 Mbps, 2 TB

Cloud Resources with Given Budget How big of a server can I support with given budget? Many project servers are sustainable on the cloud 4 Mbps, 2 TB 12 Mbps, 2 TB

Summary

Performance tradeoffs Summary 20 DG TeraFLOPS within 6 months

Summary Performance tradeoffs 20 DG TeraFLOPS within 6 months Monetary tradeoffs Client hosting After 13 days, DG more cost effective Server hosting DG server on cloud is cost-effective Best for small-medium sized projects Savings of at least 40% http://mescal.imag.fr/membres/derrick.kondo/cloud_calc.xlsx

Summary Performance tradeoffs 20 DG TeraFLOPS within 6 months Monetary tradeoffs Client hosting After 13 days, DG more cost effective Server hosting DG server on cloud is cost-effective Best for small-medium sized projects Savings of at least 40% http://mescal.imag.fr/membres/derrick.kondo/cloud_calc.xlsx BOINC server image for EC2 / Xen available http://boinc.berkeley.edu/trac/wiki/cloudserver

Future Work Clouds@home Statistical and predictive methods for ensuring host availability Lightweight mechanisms for virtual machine migration

wu flop calc bw in cloud how is cloud equiv calculated