B4 greengrid


Published on

Published in: Technology
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

B4 greengrid

  1. 1. Marcos Dias de Assunção , Jean-Patrick Gelas, Laurent Lefèvre and Anne-Cécile Orgerie INRIA RESO, Université de Lyon École Normale Supérieure 46, allée d’Italie 69364 Lyon Cedex 07 - France Grenoble, June 2011
  2. 2. <ul><li>Challenge of managing and providing resources to user applications </li></ul><ul><ul><li>Server farms, Grids, data centres and Clouds </li></ul></ul><ul><li>The Grid’5000: </li></ul><ul><ul><li>Experimental Grid composed of 9 sites distributed across France </li></ul></ul><ul><ul><li>OAR * : open-source Resource Management System (RMS) based on high-level components </li></ul></ul><ul><ul><li>Resource requests: </li></ul></ul><ul><ul><ul><li>Advance reservations System deployment with Kadeploy </li></ul></ul></ul><ul><ul><ul><li>Best-effort jobs </li></ul></ul></ul>* Nicolas Capit et al., A Batch Scheduler with High Level Components, 5th IEEE International Symposium on Cluster Computing and the Grid (CCGrid'05), pp. 776-783, May 2005
  3. 3. <ul><li>Energy consumption of IT </li></ul><ul><ul><li>CO 2 footprint of Grids and Clouds </li></ul></ul><ul><ul><li>Existing hardware, software and cooling solutions </li></ul></ul><ul><li>Evaluation of proposed solutions </li></ul><ul><li>GREEN-NET approach* </li></ul><ul><ul><li>Informing the users </li></ul></ul><ul><ul><li>Involving the users </li></ul></ul><ul><ul><li>Autonomic energy aware support </li></ul></ul>* G. Costa et al., Multi-Facet Approach to Reduce Energy Consumption in Clouds and Grids: The GREEN-NET Framework. In 1st Int. Conf. on Energy Efficient Computing and Networking, Passau, 2010.
  4. 4.
  5. 5. <ul><li>The energy sensor infrastructure : </li></ul><ul><ul><li>Users’ awareness of energy consumption </li></ul></ul><ul><ul><li>Improving the design of user applications </li></ul></ul><ul><ul><li>Energy-aware Grid middleware </li></ul></ul><ul><ul><li>Power management policies </li></ul></ul><ul><ul><li>R epository of energy consumption data </li></ul></ul><ul><li>Sensor heterogeneity </li></ul><ul><ul><li>Power meter HAMEG HM8115-2 </li></ul></ul><ul><ul><li>Omegawatt box </li></ul></ul><ul><ul><li>Library for interfacing with energy sensors </li></ul></ul>
  6. 6. <ul><li>Omegawatt box </li></ul><ul><ul><li>6 or 48 ports, communication via serial port </li></ul></ul><ul><ul><li>Deployed in three sites of Grid’5000 </li></ul></ul><ul><ul><li>Lyon, T o ulouse and Grenoble </li></ul></ul><ul><li>Client-side applications to obtain and store the energy consumption data </li></ul>* M. D. Assunção et al. The Green Grid'5000: Instrumenting and Using a Grid with Energy Sensors, 5th International Workshop on Distributed Cooperative Laboratories: Instrumenting the Grid, 2010
  7. 7. <ul><li>Periodicity of energy measurements : </li></ul><ul><ul><li>Administrators ’ requirements </li></ul></ul><ul><ul><li>Performance and energy consumption of applications </li></ul></ul><ul><ul><li>Grid middleware designers </li></ul></ul><ul><ul><li>One measurement per second for each equipment </li></ul></ul><ul><li>Data storage: </li></ul><ul><ul><li>Raw data (text files) </li></ul></ul><ul><ul><li>Last values (text files in memory) </li></ul></ul><ul><ul><li>Round-robin databases * </li></ul></ul>* RRD, http://oss.oetiker.ch/rrdtool
  8. 8.
  9. 9. <ul><li>Profiling the energy consumption of applications </li></ul>
  10. 10.
  11. 11.
  12. 12. <ul><li>Next reservation R e = (l e , n e , t e ) </li></ul><ul><li>Method 1 : At time t, the estimated start time of R e is the average of the arrival of reservations after time of the day t on: </li></ul><ul><ul><li>T he two previous days </li></ul></ul><ul><ul><li>The same weekday of the previous week </li></ul></ul><ul><ul><li>i.e. t e = 1/3 [t t,j-1 + t t,j-2 + t t,j-7 ] + t_feedback </li></ul></ul><ul><li>Method 2 : Average of characteristics of 5 previous reservations </li></ul><ul><li>Logs of advance reservation requests </li></ul>Length or duration Number of nodes Start time
  13. 13. <ul><li>U ser: always obeys the user’s demands </li></ul><ul><li>F ully-green: uses the solution that saves the most energy </li></ul><ul><li>**%-green: handles ** of requests, taken at random, with the fully-green scheme and the rest with the user policy </li></ul><ul><li>D eadline: uses the fully-green approach if it does not delay the request for more than 24h of the start time required by the user </li></ul>
  14. 14. <ul><li>Replay of Grid’5000 logs </li></ul>
  15. 15.
  16. 16. <ul><li>Analysing the impact of resource allocation policies </li></ul>~14% of the electricity consumed by the platform during ~5 months
  17. 17. <ul><li>HP Proliant G2 servers </li></ul><ul><ul><li>2.2GHz, 2 duo core CPUs per node </li></ul></ul><ul><ul><li>Xen open source 3.4.1 </li></ul></ul><ul><li>Workload </li></ul><ul><ul><li>cpuburn </li></ul></ul>* Anne-Cécile Orgerie, Marcos Dias de Assunção and Laurent Lefèvre, Energy Aware Clouds. Grids, Clouds and Virtualization, Springer-Verlag London Ltd, ISBN: 978-0-857-29048-9, Sep. 2010
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23. <ul><li>Understanding the overall infrastructure </li></ul>Marcos Dias de Assunção, Anne-Cécile Orgerie and Laurent Lefèvre, An Analysis of Power Consumption Logs from a Monitored Grid Site. IEEE/ACM International Conference on Green Computing and Communications, Hangzhou, Dec. 2010
  24. 24. <ul><li>For each reservation, evaluate the application-driven consumption </li></ul><ul><li>3.05% of the energy consumed during reservations is due to running applications or services , which is also the maximum that a CPU throttling approach could save under current conditions </li></ul>
  25. 25. <ul><li>The GREEN-NET framework </li></ul><ul><li>Energy sensors: </li></ul><ul><ul><li>Data requirements and management </li></ul></ul><ul><li>Approaches to improve the Grid middleware </li></ul><ul><li>Analysis of energy consumption logs </li></ul><ul><li>Future work </li></ul><ul><ul><li>Grid’5000 API </li></ul></ul><ul><ul><li>Network equipments and protocols </li></ul></ul><ul><ul><li>Virtualisation technologies </li></ul></ul><ul><ul><li>Repository of energy consumption data* </li></ul></ul>*http://www.ens-lyon.fr/LIP/RESO/ict-energy-logs/
  26. 26. Questions & Answers Email: assuncao at acm.org