Presentation is loading. Please wait.

Presentation is loading. Please wait.

Software Agents as Resource Managers in Grid Michał Drozdowicz, Maria Ganzha Marcin Paprzycki SRI PAS Warsaw, Poland

Similar presentations


Presentation on theme: "Software Agents as Resource Managers in Grid Michał Drozdowicz, Maria Ganzha Marcin Paprzycki SRI PAS Warsaw, Poland"— Presentation transcript:

1 Software Agents as Resource Managers in Grid Michał Drozdowicz, Maria Ganzha Marcin Paprzycki SRI PAS Warsaw, Poland http://www.ibspan.waw.pl/~paprzyck/mp/cvr/research/agent.html Mehrdad Senobari Tarbiat Modares University, Tehran, Iran Iyad Alshabani, Richard Olejnik, Mahmoud Taifour University of Sciences and Technologies of Lille Ivan Lirkov IPP BAS, Sofia, Bulgaria

2 The Grid I. Foster, C. Kesselman, The Grid: Blueprint for a New Computing Infrastructure, 1 st Edition, 1999

3 Grid – main ideas Power grid – plug the appliance and use the electric power pay for used “units of electricity” Computational Grid – plug the computer and use available resources pay for used “units of computing” Example: you do not need to buy MATLAB – you buy “MATLAB- time” you pay ?“3 cents”? for an hour of its usage Sun wants $1 per hour – is this market value!?

4 Local vs. global Grid Local Grid company / laboratory / EU-project each node / node-group → has administrator node existence “assured” (or graceful disappearance) no actual economical model involved (funny money) Global Grid collection of nodes → “seti@home” no (centralized) administration no assurances (existence / load of nodes) nodes appear and disappear node-loads change “fast” how to facilitate SLA / QoS ? economical model necessary

5 Why agents in Grid? Task of managing access to computing and data Grid resources – complex and resource consuming Little has been done on aspects of intelligence and adaptability → Next Generation Grid Expert Group’s report: “Future for European Grids: Grids and Service Oriented Knowledge Utilities - Vision and Research Directions 2010 and Beyond” Possible solution – software agents I. Foster, N. R. Jennings, C. Kesselman, Brain meets brawn: why grid and agents need each other → AAMAS ’04: Proceedings of the Third International Joint Conference on Autonomous Agents and Multiagent Systems, IEEE Computer Society, 8-1

6 Agents in Grids – key ideas B. Di Martino and O. Rana – MAGDA (Mobile AGent Distributed Application) static and mobile agents in the system static agents located within Grid nodes mobile agents visit nodes to find resources (services)‏ visits based on exchanges of messages with nodes execute a task or a part of it no economic model project moves toward SOA + choreography S. Manvi et. al. attempt at adding economic model single agent moves, negotiates, executes heavily based on mobility → can be costly

7 Our approach  assumptions Agents work in teams Each team has a team leader (local master – LMaster)‏ Incoming workers (LAgents) can join any team based on their criteria of joining Teams can accept workers based on their own criteria of acceptance Each LAgent can (if needed) play role of an LMaster Decisions about joining and accepting will utilize multi- criterial analysis Yellow-page method for matchmaking (provided by the CIC agent) → other approaches possible To support stable existence of a team each LMaster has an LMirror

8 Use Case of the system

9 AML Social Diagram

10 ONTOLOGIES

11 What we need Concepts (with properties) to describe: Resources – hardware, software, etc. SLA, QoS team work contract job execution constraints / scheduling Economical aspects of contracts Trust, reputation and reliability

12 Found approaches Traditional Grid resource description vgDL Condor Unicore Globus Toolkit → GLUE Semantic Grid description and Grid interoperability Grid resource ontology of Pernas and Dantas Grid resource ontology of Vidal Gird Interoperability Project (GRIP) → UniGrid Core Grid Ontology Agent – Grid Integration Agent Computational Grid AGIO

13 Status as of Today CoreGRID ontology is being adapted to add / clarify / modify what we believed was necessary preliminary version ready (few open issues remain) ontology can be sent to anyone interested In the system that exists at this stage we have utilized a simplistic ontology → which will need to be replaced with the above one (after (re)checking against basic use case scenarios)

14 Resource description: example :Computer :a owl:Class. :hasCPU :a owl:ObjectProperty; rdfs:range :CPU; rdfs:domain :Computer. :CPU :a owl:Class. :hasCPUFrequency :a owl:DataProperty; rdfs:comment "in GHz"; rdfs:range xsd:float; rdfs:domain :CPU. :hasCPUType :a owl:ObjectProperty; rdfs:range :CPUType; rdfs:domain :CPU. :CPUType :a owl:Class. Intel :a :CPUType. AMDAthlon :a :CPUType. :hasMemory :a owl:DatatypeProperty; rdfs:comment "in MB"; rdfs:range xsd:float; rdfs:domain :Computer. :hasUserDiskQuota :a owl:DatatypeProperty; rdfs:comment "in MB"; rdfs:range xsd:float; rdfs:domain :Computer. :LMaster :a owl:Class; :hasContactAID :a owl:ObjectProperty; rdfs:range xsd:string; rdfs:domain :LMaster. :hasWorker :a owl:ObjectProperty; rdfs:range :Worker; rdfs:domain :Computer. :Worker....

15 Description of a team :LMaster3 :hasContactAID "monster@e-plant:1099/JADE"; :hasWorker :PC2929. :PC2929 :a :Computer; :hasCPU [ a :CPU; :hasCPUType :Intel; :hasCPUFrequency "3.7"; ] ; :hasUserDiskQuota "4000"; :hasMemory "512".

16 FINDING TEAM TO DO THE JOB

17 Scenario User specifies (to the LAgent) requirements for job execution LAgent communicates with the CIC → obtains list of potential agent teams Eliminates from the list teams that are not trustworthy Utilizes FIPA Contract Net Protocol to obtain job execution proposals Eliminates teams that submitted proposals violating constraints Evaluates remaining proposals using a simple linear additive model Submits job to the selected team

18 Ontology of constraints (fragment) ### negotiation parameters ### :NegotiationSet a owl:Class. :negotiationParam a owl:ObjectProperty ; rdfs:domain :NegotiationSet ; rdfs:range NegotiationParam. :NegotiationParam a owl:Class. :paramWeight a owl:DatatypeProperty, owl:FunctionalProperty; rdfs:domain: NegotiationParam; rdfs:range xsd:float. :Cost a owl:Class; rdfs:subClassOf:NegotiationParam. :costConstraint a owl:ObjectProperty, owl:FunctionalProperty; rdfs:domain:Cost; rdfs:range:FloatConstraint. :costValue a owl:DatatypeProperty, owl:FunctionalProperty; rdfs:domain:Cost; rdfs:range xsd:float. :JobStartTime a owl:Class; rdfs:subClassOf :NegotiationParam. :jobStartTimeValue a owl:DatatypeProperty, owl:FunctionalProperty; rdfs:domain:JobStartTime; rdfs:range xsd:dateTime. ### generic constraints ### :NegotiationParamConstraint a owl:Class. :FloatConstraint a owl:Class; rdfs:subClassOf:NegotiationParamConstraint. :maxFloatValue a owl:FunctionalProperty, owl:DatatypeProperty; rdfs:domain:FloatConstraint; rdfs:range xsd:float. :minFloatValue a owl:DatatypeProperty, owl:FunctionalProperty; rdfs:domain:FloatConstraint; rdfs:range xsd:float. :TimeConstraint a owl:Class; rdfs:subClassOf:NegotiationParamConstraint. :minDateValue a owl:DatatypeProperty, owl:FunctionalProperty; rdfs:domain:TimeConstraint; rdfs:range xsd:dateTime. :maxDateValue a owl:FunctionalProperty, owl:DatatypeProperty; rdfs:domain:TimeConstraint; rdfs:range xsd:dateTime.

19 Constraint specification

20 FIPA Contract Net Protocol

21 Linear additive evaluation

22 Example

23 FINDING TEAM TO JOIN

24 Scenario User specifies (to the LAgent) requirements for team joining LAgent communicates with the CIC obtains list of potential agent teams Eliminates those that are not trustworthy Utilizes FIPA Contract Net Protocol to obtain team joining proposals Evaluates them using best price as criterion – now muticriterial analysis – in the future Confirms team joining

25 UML Sequence diagram

26 SPARQL Query PREFIX Grid: SELECT ?team WHERE { ?team Grid:needs ?machine. ?machine Grid:hasCPU ?cpu ; Grid:hasMemory ?mem ; Grid:hasQuota ?quota. FILTER ( ?cpu <= "3.7"^xsd:float ). FILTER ( ?mem <= “512"^xsd:integer ). FILTER ( ?quota <= “4000"^xsd:integer ). }

27 FIPA Contract Net Protocol

28 CFP structure and content (cfp :sender (agent-identifier :name proteus@bach:1099/JADE)‏ :receiver (agent-identifier :name zerg@chopin:1099/JADE)‏ :content ((action (agent-identifier :name zerg@chopin:1099/JADE)‏ (take-me :configuration (hardware :cpu 3.7 :memory 512 :quota 4000)‏ :conditions (condition :availability (every-day :when (period :from 00000000T23500000 :to 00000000T08150000))‏ :contract-duration +00000007T000000000))‏ :language fipa-sl0 :ontology joining-ontology :protocol fipa-contract-net )‏

29 LMaster response preparation Similar to United Devices system scoring Price for “base system” processor speed P b → P c available memory M b → M c Available disk space D b → D c B c = P c + M c + D c Final offer (  – team overhead)

30 Example

31

32 THANK YOU! QUESTIONS? Papers available at http://agentlab.swps.edu.pl Come to IMCSIT www.imcsit.org


Download ppt "Software Agents as Resource Managers in Grid Michał Drozdowicz, Maria Ganzha Marcin Paprzycki SRI PAS Warsaw, Poland"

Similar presentations


Ads by Google