Loading…

Flash Player 9 (or above) is needed to view presentations.
We have detected that you do not have it on your computer. To install it, go here.

Like this presentation? Why not share!

Workshop on Green and Environmental Computing - Helsinki

on

  • 1,273 views

 

Statistics

Views

Total Views
1,273
Views on SlideShare
939
Embed Views
334

Actions

Likes
0
Downloads
39
Comments
0

3 Embeds 334

http://www.jorditorres.org 332
http://webcache.googleusercontent.com 1
http://www.google.es 1

Accessibility

Categories

Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

Workshop on Green and Environmental Computing - Helsinki Workshop on Green and Environmental Computing - Helsinki Presentation Transcript

  • Workshop on Green and Environmental Computing p g 27th – 28th October 2010 CSC – IT Center for Science Espoo, Finland Middleware research Middl h for green Data Centers Prof. Jordi Torres Barcelona Supercomputing Center (BSC) UPC – Barcelona Tech Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 1
  • Contents Motivation for reduction of energy consumption What can we do for DC power efficiency? Power aware Resource Management Research at a Glance Conclusions Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 2
  • The motivations are well-known by all of you Impact of computing on the environment: ICT industry produces roughly 2 to 3 percent of global GHG emissions. Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 3 View slide
  • And we can’t forget … g Opex Data Center 5 years (%) Calculation for avg density 7kw/rack Data courtesy of Infrastructure amortization counted as Opex AST IT Infrastructures HW and SW excluded (www.AST-global.com) Data Center Infrastructure valued as Opex Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 4 View slide
  • Goal: Reduce DC energy use gy Has become a first-order objective in the design and operation of the Data Centers Source: EPA report to Congress 2007 How can we reduce it? Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 5
  • Contents Motivation for reduction of energy consumption What can we do for DC power efficiency? Power aware Resource Management Research at a Glance Conclusions Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 6
  • What can we do for DC power efficiency? Can we Minimize losses and thermal/cooling overheads in a Data Center? Example: Reducing cooling costs “Best Practices to save energy at Marenostrum” Source: “El reto operacional de dirigir el tercer supercomputador más grande de Europa” Sergi Girona, Director Operaciones BSC-CNS, 2009. Foto: BSC Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 7
  • Best practice at Marenostrum Problem measured: to much under-floor pressure Test: Move some floor tiles Benefits observed: – Improvement of AC equipment performance – Improved the rack- p bottom temperature Foto: BSC Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 8
  • Best practice at Marenostrum Substituting Floor Tiles – Composite • 20% opening – Metallic tile • 40% opening Benefits observed: B fit b d – Less working pressure for the Cooling components – All bladecenters reduced by 2º C – Cold barrier that prevents the reflux of hot air Foto: BSC Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 9
  • Best practice at Marenostrum Problem measured : not all the Racks have the same temperature Proposal: Methacrylate screens installed in front of each rack Benefits observed: – All BladeCenters Rack has an equal t l temperature +/- 1ºC t / – BladeCenter fan speed reduced Foto: BSC Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 10
  • Best practice at Marenostrum Results (including other improvements): reduction of 10% of power consumption (and CO2) Image courtesy of UPC Marenostrum power consumption: approx. 1.2 approx 1 2 Mwats aprox 1 100 000 €/year 1.100.000 Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 11
  • Decreasing DC power consumption How can we decrease DC power consumption without a concomitant decrease in the IT load? Can we move the Marenostrum in a better location? e.g. more efficient cooling due to climate conditions Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 12
  • Can we put a DC in a better location? I containerized marenostrum and it has traveled with me Helsinki PUE = 1,073 Data courtesy of Barcelona AST IT Infrastructures PUE = 1,113 (www.AST-global.com) Estimated ~3% of energy could be saved Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 13
  • What else can we do? Maximize work done per watt Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 14
  • What else can we do? Maximize work done per watt Energy consumption is not only determined by the efficiency of the physical resources but it is also dependent on the resource management Intelligent management of resources may lead to significant reduction of the energy consumption b i ifi d i f h i by meeting the performance requirements Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 15
  • Contents Motivation for reduction of energy consumption What can we do for DC power efficiency? Power aware Resource Management Research at a Glance Conclusions Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 16
  • Power aware Resource Management Different layers in a Data Center? Data Center Infrastructure Applications System Hardware Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 17
  • Different layers in a Data Center y Who is responsible for managing resources? Data Center Infrastructure … Web Services W bS i Aplication Server Applications Web server Java Virtual Machine System Operating System Virtualization layer Hardware … MIDDLEWARE Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 18
  • Power/Energy Aware Resource Management at Middleware level Challenge: – Shift the paradigm from “time to solution" to "kWh to solution” How?: – Introducing intelligent self-management techniques – Adding energy efficiency parameter to the list of critical operating parameters to control that already including availability, reliability, performance, … il bilit li bilit f Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 19
  • Contents Motivation for reduction of energy consumption What can we do for DC power efficiency? Power aware Resource Management Research at a Glance Conclusions Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 20
  • Reseach at a Glance Today the most common techniques/control knobs of the middleware are related to: – consolidation and virtualization technology, – turning on/off policies, – new processors, processors – dynamic voltage and frequency scaling, – hybrid architectures on datacenters, y , – federating clouds – etc. Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 21
  • Workload Consolidation Rationalize and reduce operational costs: server consolidation – implies combining workloads from separate applications into a smaller number of systems if there are different peak times C1 ement ement C2 rce rce Require Require Resour Resour Time Time C1 quirement C2 source Req Res Time Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 22
  • Virtualization Technologies Consolidation uses virtualization to share the resources – Virtualization software divides a physical server into isolated virtual environments, enabling organizations to run multiple applications or OS on a single server server. App 1 App 2 App 3 Operating System Operating System Virtual Machine Virtual Machine App 1 App 2 App 3 ...... CPU Mem CPU Mem Operating System Physical Physical Machine CPU CPU CPU CPU Mem Machine CPU CPU CPU CPU Mem The energy savings can be up to 20% - 70% Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 23
  • Using multi-core processors Middleware allow that applications benefit from new multi-core designs beyond the traditional parallelism Example: Exploitation of Accelerators in memory-bound memory bound Applications using memory compression techniques App 1 App 2 App 3 Operating System Operating System Virtual Machine Virtual Machine ...... CPU Mem CPU Mem Mem Physical Machine CPU CPU CPU CPU Mem For F memory-bound appl. the energy savings can be up to 25% b d l th i b t Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 24
  • Use low-power processors Low-power processors vs high-performance processors Good approach for transactional workloads (webs) – Experiments at BSC: Reduction of 70% Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 25
  • Turn on/off policies However the numerical applications do not behave like this Experiments at BSC with 4 HPC tasks: • 1 Xeon * 1 hour 317 Watts • 2 Atom * 5 hours 398 Watts – The most energy-efficient approach in this environment is to run jobs very fast and then turn off the power of the system. system Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 26
  • Hybrid hardware y But in general the workload in data centers is heterogeneous – F HPC, Xeons work more efficiently th At For HPC X k ffi i tl than Atoms – For Transactional, Atoms work more efficiently than Xeons Transactional Therefore, why , y not build the two toget e together? Courtesy of Giovanni Leonardi (Mataró-Sicilia) Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 27
  • Hybrid hardware y Better tradeoff in terms of energy and revenue – Experiments at BSC with heterogeneous workload: Reduction of 25% Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 28
  • In the Cloud Era Resource Management in a Cloud Federation g – Distribution of workload across geographically distributed DC – Consolidation at DC level (and turn off a whole DC) – Possibility to reallocate the workload to a place where energy or cooling is cheaper e.g. solar energy during daytime across different time zones, efficient cooling due to climate conditions, etc. conditions etc – Etc Etc. Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 29
  • In the Cloud Era Outsourcing/Insourcing strategies g g g The energy savings can be up to 10%? 30? 50%? Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 30
  • Conclusions The energy savings can be up to 20% - 70% Reduction of 25% Reduction of 70% For memory-bound appl the energy savings can be up to 25% appl. The energy savings can be up to 10%? 30? 50%? Resource management at Middleware level will play an important role in the reduction of energy consumption Further research is required! Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 31
  • Conclusions Future research directions in resource management – Development of energy-efficient resource management approaches for multi-core CPUs due to the wide adoption of these. – Development of intelligent techniques to manage resources efficienty. – Efficient workload distribution across geographically distributed DC. – SLA negotiation between resource providers and users that includes control over power consumption and/or CO2 emissions. – Etc. Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 32
  • Workshop on Green and Environmental Computing p g 27th – 28th October 2010 CSC – IT Center for Science Espoo, Finland Middleware research Middl h for green Data Centers Prof. Jordi Torres Barcelona Supercomputing Center (BSC) UPC – Barcelona Tech Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 33
  • Workshop on Green and Environmental Computing p g 27th – 28th October 2010 CSC – IT Center for Science Espoo, Finland About the author: Jordi Torres has a Master degree in Computer Science from the Technical University of Catalonia (UPC Barcelona Tech, 1988) and also holds a Ph. D. y ( ) from the same institution (Best UPC Computer Science Thesis Award, 1993). Currently he is a full-time professor in the Computer Architecture Department at UPC and is a manager for the Autonomic Systems and eBusiness Platforms research line at Barcelona Supercomputing Center (BSC). His current principal interest as a researcher is to make IT resources more efficient i order t obtain more sustainable IT and t f ffi i t in d to bt i t i bl d to focus on th resource the management needs of modern distributed and parallel cloud computing environments. He has worked on a number of EU and industrial research and development projects. He has more than a hundred publications in the area and was i d involved i several conferences i th fi ld of di t ib t d and l d in l f in the field f distributed d parallel systems. He has been Vice-dean of Institutional Relations at the Computer Science School, and currently he is a member of the University Senate and member , y y of the Board of Governors. He can be reached at www.JordiTorres.org Jordi Torres – Barcelona Supercomputing Center (BSC) 2010 e-InfraNet Workshop on Green and Environmental Computing 34