A Distributed Resource Management Architecture that supports Advance Reservation and Co-Allocation

1. A Distributed Resource Management Architecture that supports Advance Reservation and Co-Allocation Ian Foster, Carl Kesselman, Craig Lee, Bob Lindell, Klara Nahrstedt & Alain Roy Reporter：S.Y.Chen

2. Abstract • Primary Requirement : End-to-End QoS • Issues pertaining to Resources. • Dynamic Discovery • Reservation (Advance or Immediate) • Heterogeneous Resources • Independently Controlled and Administered • How GARA addresses these issues. • A prototype implementation and results • It propose the Grid Architecture for Reservation and allocation (GARA) and describe a prototype GARA that supports three different resource types S.Y. Chen

3. Outline • Introduction • GARA Architecture Design • Co-Reservation / Allocation Agents • GARA Application Programming Interfaces • GARA Implementation • Results • Conclusions S.Y. Chen

4. Introduction • Configuration of Resources • Resource Discovery • Resource Selection • Resource Reservation • Advance Reservation (e.g. Airline Ticket) • Immediate Reservation (e.g RSVP) • Resource Allocation S.Y. Chen

5. Introduction (cont.) • Deployed systems lack support for advanced reservation • Complex collections of resources • Varying types of resources. • Resources present in different administrative domains. S.Y. Chen

6. GARA Architecture Design • The Globus Resource Management Architecture (GRMA) • Globus Architecture for Reservation and Allocation (GARA) S.Y. Chen

7. GARA Architecture Design - GRMA • Provides dedicated access to collection of computers in heterogeneous distributed systems. • The architecture consists of three main components • Information Service • Various types of Co-Allocation Agents • Local Resource Managers S.Y. Chen

8. GARA Architecture Design - GRMA (cont.) • The Globus Resource Management Architecture S.Y. Chen

9. GARA Architecture Design - GARA • Globus Architecture for Reservation and Allocation S.Y. Chen

10. Co-reservation / Allocation agents S.Y. Chen

11. GARA Application Programming Interfaces • CreateReservation • CreateObject • CancelReservation • CancelObject • RegisterCallBack • ModifyReservation S.Y. Chen

12. GARA API (cont.) S.Y. Chen

13. GARA Implementation • The structure of a particular LRAM implementation depends upon the nature of the local resource management services. (e.g., the “scheduler”) Three major cases can be distinguished: • 1. If the scheduler provides appropriate advance reservation support, then an LRAM for that resource can pass advance reservation requests directly to the scheduler. S.Y. Chen

14. GARA Implementation (cont.) • 2.Otherwise, we have to deal with two cases: • (a) If an LRAM associated with a resource has total control over the resource (i.e., all object creation calls must pass via the LRAM), then the LRAM can use a slot manager to implement advance reservations for that resource. • (b) Otherwise, only probabilistic advance reservations can be supported. S.Y. Chen

15. GARA Implementation (cont.) • A slot table keeps track of current allocations and future reservations for a resource S.Y. Chen

16. GARA Implementation (cont.) • The three GARA resource managers constructed S.Y. Chen

17. Results • Sun Ultra-2 workstations • Solaris 2.6 • 100 Mb/s Ethernet S.Y. Chen

18. Conclusions • GARA solves the problem of achieving end-to-end QoS guarantees across heterogeneous collections of shared resources. • Co-reservation & co-allocation agents discover, reserve and allocate resources. • Cost of GARA mechanisms are not large when compared to underlying resource management operations. S.Y. Chen