Cracow Grid Workshop October 16, 2006 Economic Virtualization of ICT Infrastructures Jochen Stoesser - PowerPoint PPT Presentation

About This Presentation
Title:

Cracow Grid Workshop October 16, 2006 Economic Virtualization of ICT Infrastructures Jochen Stoesser

Description:

Jochen Stoesser, Arun Anandasivam, Nikolay Borissov, Dirk Neumann ... mailto:stoesser_at_iism.uka.de. IISM Information & Market Engineering. Universit t Karlsruhe (TH) ... – PowerPoint PPT presentation

Number of Views:35
Avg rating:3.0/5.0
Slides: 15
Provided by: cyfr
Category:

less

Transcript and Presenter's Notes

Title: Cracow Grid Workshop October 16, 2006 Economic Virtualization of ICT Infrastructures Jochen Stoesser


1
Cracow Grid WorkshopOctober 16, 2006Economic
Virtualization of ICT InfrastructuresJochen
Stoesser, Arun Anandasivam, Nikolay Borissov,
Dirk NeumannInstitute of Information Systems and
Management (IISM)University of Karlsruhe, Germany
2
Agenda
  • Requirements for efficient resource allocation in
    the Grid
  • Drawbacks of current approaches to pricing and
    scheduling
  • Towards market-based resource allocation
    combinatorial double auctions

3
Requirements for efficient resource allocation in
the Grid
  • Basic requirements
  • Demand for and supply of resources highly
    dynamic
  • Users (providers) require (offer) combinations of
    resources
  • memory, compute, memory, compute,
    application,
  • If demand higher than supply of resources award
    resources to requesters who value these resources
    the most.

4
Current approaches to scheduling and pricing
  • Subscription model
  • buyers pay a flat rate on a periodic
  • basis to use a resource
  • inflexible
  • Metered model
  • Pay as you go, e.g. Amazons Compute Cloud, SUN
  • users are charged on the basis of actual usage
  • prices are (temporarily) determined by providers
    only ? inflexible
  • Central scheduling
  • who shares with whom at what time
  • if demand exceeds supply, allocation of resources
    becomes inefficient

q
actual demand/supply
service level
t
5
Objective market-based resource allocation
Dynamic and efficient allocation of scarce
resources ? Typical application domain of markets
Buyya et al. 2003
  • Objective Development of an Open Grid Market
    that
  • is flexible and dynamic,
  • provides economic incentives for resource
    provision and
  • achieves an efficient allocation of resources.

Grid research projects at the IISM
SESAM
Billing the Grid
6
Market-based resource allocation Conventional
auction mechanisms
  • Conventional single-sided auctions (cf. eBay)
    and double auctions
  • (cf. stock exchanges) not efficient
  • Split of supply and demand across multiple
    markets
  • Exposure risk requester might only obtain a
    subset of the required resources

resource provider 2 application
resource provider 1 memory, compute
Deal
No deal
resource requester memory, compute, application
7
Market-based resource allocation Combinatorial
double auction
  • single bid for a combination of resources
  • in an aggregated market
  • actors can bid for bundles of resources, e.g.
    memory and compute
  • logical links between bids, e.g. XOR bids
  • Bapna et al., forthcoming de Vries et al. 2003
    Schnizler et al. forthcoming

m resource providers

memory, compute, application

n resource requesters
8
Market-based resource allocation Combinatorial
double auction (contd)
  • Problems
  • Combinatorial assignment problem
  • hard to model
  • computationally extremely demanding
  • can be reduced to multi-dimensional knapsack
    problem ? NP-hard
  • Trade-off between economic requirements required
    Parkes et al., 2001
  • Incentive compatible
  • Efficient allocation
  • Budget balanced
  • Need for additional attributes SLAs, location,
    time constraints,

9
Contribution
  • Discussion of related approaches
  • Parameterisation of the design space
  • derivation of use cases
  • market engineering approach

10
Contribution (contd)
  • Formalization of the bidding language and the
    winner determination problem for sample use cases

Efficiency criterion
Constraints encode the use case

11
Thank you for your attention!
Contact Information
Jochen Stoesser mailtostoesser_at_iism.uka.de IISM
Information Market Engineering Universität
Karlsruhe (TH) Englerstr. 14, 76 131 Karlsruhe,
Germany www.iism.uni-karlsruhe.de www.sorma-projec
t.eu
12
References
  • Rajkumar Buyya, David Abramson, Srikumar
    Venugopal. The Grid Economy. Proceedings of the
    IEEE, Vol. 93, No. 3, March 2005.
  • David C. Parkes, Jayant Kalagnanam, Marta Eso.
    Achieving budget-balance with vickrey-based
    payment schemes in exchanges. In IJCAI, pp.
    1161-1168, 2001.
  • Ravi Bapna, Sanjunkta Das, Robert Garfinkel,
    Jan Stallaert. A market design for grid
    computing. INFORMS Journal of Computing,
    forthcoming.
  • Sven de Vries, Rakesh V. Vohra. Combinatorial
    Auctions A Survey. INFORMS Journal on Computing,
    Vol. 15, No. 3, pp. 284-309, Summer 2003.
  • Björn Schnizler, Dirk Neumann, Daniel Veit,
    Christof Weinhardt. Trading Grid Services A
    Multi-attribute Combinatorial Approach. European
    Journal of Operations Research, forthcoming.

13
Combinatorial double auction Formalisation
  • Winner determination problem (WDP)


14
Parametrization of the design space
Write a Comment
User Comments (0)
About PowerShow.com