Utility-based multi-category quality-of-service negotiation in distributed systems

• US 6,154,778 A
• Filed: 05/19/1998
• Issued: 11/28/2000
• Est. Priority Date: 05/19/1998
• Status: Expired due to Term

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Company (HP Inc.)
• Inventors: Seetharaman, Aparna, Kirshenbaum, Evan R., Koistinen, Jari
• Primary Examiner(s): Burgess, Glenton B.
• Assistant Examiner(s): Salad, Abdullahi E.
• Application Number:
  US09/081,265
• Time in Patent Office: 924 Days
• Field of Search: 709/228, 709/203, 709/227, 709/240, 709/241, 709/223, 709/224, 709/239, 370/395, 370/409, 370/230, 370/465
• US Class Current: 709/228
• CPC Class Codes: H04L 2012/5631 Resource management and allocation
• CPC Class Codes: H04L 2012/5646 Cell characteristics, e.g. loss, delay, jitter, sequence integrity
• CPC Class Codes: H04Q 11/0478 Provisions for broadband connections

Abstract

In a distributed system, a method and system for negotiating a multi-category Quality-of-Service (QoS) agreement between a client and a server includes a client agent enabled to calculate an expected utility to a client of multiple multi-category QoS specifications. The client agent obtains the QoS specifications by transmitting a QoS specification request to a server agent or a broker. The expected utility calculation, based on a probabilistic estimate of QoS levels included in the QoS specifications, enables the client agent to distinguish the QoS specifications of greater value from those of lesser value. The client agent selects at least one of the QoS specifications to be included into an offer for a QoS agreement based on the expected utility calculation. In a preferred embodiment, the client agent selects the QoS specifications determined to be most valuable to the client. The offer is transmitted to the server agent to request a service provided by a server at QoS levels represented by the selected QoS specifications. After transmitting the offer, the client monitors a connection to the server agent for either an acceptance, a rejection, or a counteroffer to the offer. Communication between the client agent and the server agent conforms to a negotiation protocol which provides a set of rules for transmission of negotiation messages.

Claims

1. A method for negotiating a multi-category quality-of-service (QoS) agreement between a client and a server in a system comprising the steps of:

establishing a communications link having a client-side of said negotiating and server-side;

generating a plurality of multi-category QoS specifications, each said multi-category QoS specification being representative of a probabilistic estimate of QoS levels for a service available via said server, each category of said QoS specifications representing a different QoS-related parameter;

transmitting said plurality of multi-category QoS specifications over said communications link from said server-side to said client-side;

selecting at least one of said multi-category QoS specifications to be included in an offer for a QoS agreement;

identifying each said selected multi-category QoS specification as part of said offer, said offer requesting a service provided by said server at said probabilistic estimate of said QoS levels represented by said each selected multi-category QoS specification;

transmitting said offer over said communications link from said client-side to said server-side; and

monitoring said communications link at said client-side for a response to said offer indicating one of an acceptance, a rejection, and a counteroffer to said offer.

2. The method of claim 1 further comprising the step of calculating an expected utility to said client for each said QoS specification, including calculating expected utilities of categories of said each multi-category QoS specification received via said communications link, said calculation of said expected utilities being performed at said client-side of said communications link and being based on said probabilistic estimate of said QoS levels available via said server, and further, wherein said step of selecting said at least one multi-category QoS specification includes basing said selection on said calculation of said expected utility.

3. The method of claim 2 wherein expected utility is specific to relative significance to said client and wherein said calculating step for each said multi-category QoS specification includes a hierarchical approach of:

establishing low level expected utilities for each of a plurality of dimensions associated with different categories of said multi-category QoS specifications, each dimension being representative of one of a qualitative and a quantitative QoS-related parameter of a category with which said each dimension is associated;

establishing an intermediate level expected utility for each said category based upon said low level expected utilities established for dimensions associated with said each category, said intermediate level expected utilities being said expected utilities of said categories; and

establishing a high level expected utility for each said multi-category QoS specification based upon said intermediate level expected utilities established for said each multi-category QoS specification, said high level expected utility being said expected utility of said each multi-category QoS specification.

4. The method of claim 2 further comprising the step of:

determining whether said expected utility of each said multi-category QoS specification satisfies a preselected minimum expected utility threshold; and

if an expected utility of a particular multi-category QoS specification does not satisfy said minimum expected utility threshold, disabling consideration of said particular multi-category QoS specification from inclusion into said offer.

5. The method of claim 2 further comprising the step of referencing a constraint profile which defines minimum requirements of said client to determine whether to perform said step of calculating said expected utility.

6. The method of claim 2 wherein said step of calculating said expected utility includes filtering said each multi-category QoS specification utilizing a trust function representative of a level of trust said client has with respect to said server, said trust level being indicative of a degree to which said server has satisfactorily performed under terms of past QoS agreements.

7. The method of claim 1 further comprising the steps of:

generating a preliminary multi-category QoS specification utilizing an application-associated with said server; and

filtering said preliminary multi-category QoS specification to modify said preliminary multi-category QoS specification according to a capability of said system to support preliminary QoS levels represented by said preliminary multi-category QoS specification, said filtering of said preliminary multi-category QoS specification generating said at least one multi-category QoS specification.

8. The method of claim 1 further comprising the steps of:

receiving said offer that includes said at least one selected multi-category QoS specification;

determining whether said system is capable of supporting a QoS level associated with said selected multi-category QoS specification, said determination being based at least partially on previous commitments assigned to said system; and

deciding whether to accept said offer on behalf of said server based at least partially on said determination of whether said system is capable of supporting said QoS level.

9. A system for enabling multi-category QoS negotiation toward a multi-category QoS agreement between a server and a client in a system comprising:

a server agent comprising;

a) means for establishing multi-category QoS specifications representative of probabilistic estimates of QoS levels associated with alternative services provided by said server;

b) first means for transmitting a plurality of said multi-category QoS specifications in response to receiving a multi-category QoS specification request for communication involving said client; and

c) means for determining whether to enable said services based upon multi-category QoS offers received on behalf of said server; and

a client agent comprising;

a) means, responsive to said first means to receive said plurality of multi-category QoS specifications, for selecting one of said multi-category QoS specifications for identification to said server agent on behalf of said client; and

b) second means for identifying said selected multi-category QoS specification to said determining means of said server agent as at least a portion of a selected QoS offer.

10. The system of claim 9 wherein said client agent further includes means, connected to said first means and said selecting means, for calculating an expected utility to said client for each of said multi-category QoS specifications based on said probabilistic estimates of said QoS levels, and further, wherein said selecting means is responsive to said expected utility calculation by said calculating means.

11. The system of claim 9 wherein said probabilistic estimates of said QoS specifications are expressed for said categories of said multi-category QoS specifications in the absence of a linear combination of said probabilistic estimates.

12. The system of claim 10 wherein said calculating means of said client agent includes a client-agent trust filter configured to modify data within said multi-category QoS specification according to a first trust function, said first trust function being representative of a level of trust possessed by said client agent with respect to said server agent based on an extent to which QoS agreements negotiated by said server agent have been satisfactorily performed in the past, said determining means of said server agent including a server-agent trust filter configured to modify data within said selected multi-category QoS offer according to a second trust function, said second trust function being representative of a level of trust possessed by said server agent with respect to said client agent based on an extent to which QoS agreements negotiated by said client agent have been satisfactorily performed in the past.

13. The system of claim 10 wherein said calculating means of said client agent includes an expected default provision function enabled to calculate expected utilities for dimensions of said categories not included in said multi-category QoS specifications.

14. The system of claim 9 wherein said server agent further includes:

means, connected to said determining means of said server agent, for ascertaining whether sufficient resources are available to said server to support resource requirements of said selected multi-category QoS offer; and

means, responsive to said determining means and said ascertaining means, for selecting a second multi-category QoS specification for inclusion into a counteroffer in response to a determination that said resource requirements are unavailable to said server.

15. The system of claim 9 wherein said establishing means of said server agent includes an application enabled to generate a preliminary multi-category QoS specification, said establishing means further including a plurality of filters associated with resources of said system supporting said service provided by said server, said filters being configured to modify said preliminary multi-category QoS specification to reflect capabilities of said resources to support said service, said modification of said preliminary multi-category QoS specification resulting in at least one of said multi-category QoS specifications.

16. The system of claim 9 wherein said first means for transmitting and said second means for identifying both include a negotiation protocol having rules for transmission of messages during said multi-category QoS negotiation.

17. A method for negotiation of a multi-category QoS deal between a server and a client in a distributed system comprising the steps of:

generating an initial multi-category QoS specification representative of probabilistic estimates of QoS levels available via said server based on preferences by a server application;

filtering said initial multi-category QoS specification according to capabilities of resources of said distributed system to support QoS levels, said filtering producing a plurality of multi-category QoS specifications, each category of said multi-category QoS specifications being representative of a parameter of said multi-category QoS specification;

transmitting said plurality of multi-category QoS specifications to a client agent in response to a request via a communications link;

upon receiving said plurality of multi-category QoS specifications at said client agent, calculating an expected utility at said client agent for each of said plurality of multi-category QoS specifications based on said probability estimates of said QoS levels;

formulating an offer including at least one of said multi-category QoS specifications in response to said calculations of expected utilities;

identifying said offer to a server agent; and

upon receiving said offer at said server agent, determining whether to accept said offer, including utilizing a first trust function that is representative of a level of trust said server agent possesses with regard to said client agent based on an extent to which QoS agreements negotiated by said client agent have been satisfactorily performed previously, said calculating step including utilizing a second trust function in calculating said expected utility of said multi-category QoS specification, said second trust function being representative of a level of trust said client agent possesses with regard to said server agent based on an extent to which QoS agreements negotiated by said server agent have been satisfactorily performed previously.

18. The method of claim 17 wherein said determining step includes ranking said at least one selected multi-category QoS specification according to an expected utility to said server of each of said at least one selected multi-category QoS specification in said offer.

19. The method of claim 17 wherein said filtering step includes producing said plurality of multi-category QoS specifications such that each multi-category QoS specification includes a client behavior mode with a probabilistic estimate of client behavior.