37
Introduction Algorithm Design Evaluation Results Conclusion Modeling and Optimization of Resource Allocation in Cloud PhD Thesis Progress – Second Report Atakan Aral Thesis Advisor: Asst. Prof. Dr. Tolga Ovatman Istanbul Technical University – Department of Computer Engineering June 22, 2015 Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

Embed Size (px)

Citation preview

Page 1: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Modeling and Optimization of Resource Allocation in CloudPhD Thesis Progress – Second Report

Atakan Aral

Thesis Advisor: Asst. Prof. Dr. Tolga Ovatman

Istanbul Technical University – Department of Computer Engineering

June 22, 2015

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 2: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Outline

1 IntroductionPreliminary InformationContribution to the ThesisTime Plan

2 Algorithm Design

3 Evaluation

4 Results

5 Conclusion

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 3: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Preliminary InformationContribution to the ThesisTime Plan

Outline

1 IntroductionPreliminary InformationContribution to the ThesisTime Plan

2 Algorithm Design

3 Evaluation

4 Results

5 Conclusion

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 4: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Preliminary InformationContribution to the ThesisTime Plan

Outline

1 IntroductionPreliminary InformationContribution to the ThesisTime Plan

2 Algorithm Design

3 Evaluation

4 Results

5 Conclusion

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 5: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Preliminary InformationContribution to the ThesisTime Plan

Federated Cloud

Definition

Mechanisms and policies for scaling hosted services across multiple,geographically distributed data centers and dynamically coordinating loaddistribution among these data centers.

Aims an open and online cloud economy in which providers:operate as parts of a market driven resource leasing federation;can dynamically partner with each other to create a seemingly infinite pool of ITresources.

While users of cloud infrastructure:avoid vendor lock-in and can easily hybridize their private data center;can scale VMs across multiple IaaS providers in different geo-locations.

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 6: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Preliminary InformationContribution to the ThesisTime Plan

Distributed VMs

Opportunities:Available mechanisms and policies such as Federated Cloud;Very high speed inter-DC communication technologies such as optical fiber;Programming models that minimize size of data flow between nodes such asMapReduce

Advantages:fault tolerancevendor independencecloser proximity to user basecost benefits

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 7: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Preliminary InformationContribution to the ThesisTime Plan

Distributed VMs

VM Placement Risks:Cooperating VMs on distant DCs;VMs far away from their user base;VMs placed without considering different pricing strategies of vendors

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 8: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Preliminary InformationContribution to the ThesisTime Plan

Outline

1 IntroductionPreliminary InformationContribution to the ThesisTime Plan

2 Algorithm Design

3 Evaluation

4 Results

5 Conclusion

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 9: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Preliminary InformationContribution to the ThesisTime Plan

Algorithm Design and Implementation

Suggested Topology Based Matching (TBM) algorithm employs a graphtheoretical approach in combination with some heuristics.Incremental developmentRe-evaluation after each new improvement

to compare against baselinesto detect bottlenecks and other problems

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 10: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Preliminary InformationContribution to the ThesisTime Plan

Evaluation

Bandwidth modelingCost modelingLoad modelingEvaluation variables

7 baseline methods, 12 performance criteria, 4 variables

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 11: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Preliminary InformationContribution to the ThesisTime Plan

Documentation

A conference paper to be presented, another journal paper being writtenApplication for TUBITAK 1002 - Short Term R&D Funding ProgramBatch evaluation process which generates and logs results and charts foreach runRevision control and documentation (https://github.com/atary/RalloCloud/)

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 12: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Preliminary InformationContribution to the ThesisTime Plan

Outline

1 IntroductionPreliminary InformationContribution to the ThesisTime Plan

2 Algorithm Design

3 Evaluation

4 Results

5 Conclusion

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 13: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Preliminary InformationContribution to the ThesisTime Plan

Gantt Chart

2015

1 2 3 4 5 6

Algorithm Design

Implementation

Evaluation

Modification

Documentation

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 14: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Outline

1 IntroductionPreliminary InformationContribution to the ThesisTime Plan

2 Algorithm Design

3 Evaluation

4 Results

5 Conclusion

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 15: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Objective

To decrease deployment delay (by placing VMs close to the broker)To decrease communication delay (by placing connected VMs to theneighbour data centers)To reduce resource costs (by balancing load and avoiding overload in any DC)

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 16: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

UML Activity Diagram

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 17: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Subgraph Matching

Search space is all possible injective matchings from the set of pattern nodesto the set of target nodes.Systematically explore the search space:

Start from an empty matchingExtend the partial matching by matching a non matched pattern node to a nonmatched target nodeBacktrack if some edges are not matchedRepeat until all pattern nodes are matched (success) or all matchings arealready explored (fail).

Filters are necessary to reduce the search space by pruning branches that donot contain solutions.

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 18: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

LAD Filtering

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 19: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

LAD Filtering

D1 = D3 = D5 = D6 = A,B,C,D,E ,F ,G

D2 = D4 = A,B,D

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 20: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Outline

1 IntroductionPreliminary InformationContribution to the ThesisTime Plan

2 Algorithm Design

3 Evaluation

4 Results

5 Conclusion

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 21: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Bandwidth Modeling

Bandwidth capacities are modeled not in links but in DCs.When a link is utilized, same amount of bandwidth is reduced from the DCs inboth sides of the link.More generally, bandwidth capacities of all the nodes that are on the shortestpath are utilized.Bandwidth request between two VMs is nonbifurcated. (No path-splitting)

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 22: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Bandwidth Modeling

VM1 VM3VM2

VM1

VM2VM3

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 23: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Cost Modeling

1 Fixed pricing based on memory, bandwidth and duration.2 Dynamic pricing via Yield management

Increase the price of the resource that is running low in a DCCost = minCost + (maxCost − minCost) ∗ Util

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 24: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Load Modeling

Number of Clusters Based on the population densityaround each location. Range: 1:16

Number of VMs Based on Poisson distribution: λ = 3

Cluster Topologies Either linear or completeArrival Times Uniform random in the range [0,50)

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 25: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Evaluation Variables

VM memory request Data center memory capacity is 64x and each VM requiresmemory allocation between 1x and 8x

Link bandwidth request Available bandwidth in each link is 80y and bandwidthallocation to/from other VMs is between 1y and 8y .

Minimum number of requests Average number of requests from the leastpopulated location is in the range [2,16] depending on this variable.

VM network intensity Ratio of local computation and inter-VM communication isbetween 3 and 1/3.

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 26: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Outline

1 IntroductionPreliminary InformationContribution to the ThesisTime Plan

2 Algorithm Design

3 Evaluation

4 Results

5 Conclusion

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 27: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Performance Criteria and Baseline Heuristics

Arbitrary Next-fit (ANF)Load Balancing (LBG)Random Choice (RAN)Latency based Next-fit (LNF)

VM Deployment Latency (Seconds)VM Communication Latency (Seconds)Task Completion Time (Hours)Throughput (MIPS)Rejection Rate (%)Cost ($)

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 28: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

VM memory request

0,00

50,00

100,00

150,00

200,00

250,00

300,00

1 2 3 4 5 6 7 8

VM

De

plo

yme

nt

La

ten

cy(S

eco

nd

s)

VM RAM

ANF LBG RAN TBF LNF

0,0

0,5

1,0

1,5

2,0

2,5

3,0

1 2 3 4 5 6 7 8

VM

Com

mu

nic

atio

nL

ate

ncy

(Seco

nd

s)

VM RAM

ANF LBG RAN TBF LNF

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 29: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

VM memory request

0

50

100

150

200

250

300

350

400

1 2 3 4 5 6 7 8

Ta

skC

om

ple

tion

Tim

e (

Hou

rs)

VM RAM

ANF LBG RAN TBF LNF

0

500

1000

1500

2000

2500

3000

1 2 3 4 5 6 7 8

Th

roug

hp

ut

(MIP

S)

VM RAM

ANF LBG RAN TBF LNF

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 30: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

VM memory request

0

10

20

30

40

50

60

70

80

90

100

1 2 3 4 5 6 7 8

Re

ject

ion

Ra

te (

%)

VM RAM

ANF LBG RAN TBF LNF

0

10000

20000

30000

40000

50000

60000

1 2 3 4 5 6 7 8

Cost

($)

VM RAM

ANF LBG RAN TBF LNF

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 31: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Link bandwidth request

0,0

0,5

1,0

1,5

2,0

2,5

1 2 3 4 5 6 7 8

VM

Com

mu

nic

atio

nL

ate

ncy

(Seco

nd

s)

Link BW

ANF LBG RAN TBF LNF

0

10000

20000

30000

40000

50000

60000

1 2 3 4 5 6 7 8

Cost

($)

Link BW

ANF LBG RAN TBF LNF

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 32: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Minimum number of requests

0

500

1000

1500

2000

2500

1 2 3 4 5 6 7 8

Th

roug

hp

ut

(MIP

S)

Number of Requests

ANF LBG RAN TBF LNF

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 33: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

VM network intensity

0

50

100

150

200

250

300

350

1 2 3 4 5 6 7 8

Ta

skC

om

ple

tion

Tim

e (

Hou

rs)

Network intensity

ANF LBG RAN TBF LNF

0

10000

20000

30000

40000

50000

60000

70000

80000

90000

1 2 3 4 5 6 7 8

Cost

($)

Network intensity

ANF LBG RAN TBF LNF

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 34: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Outline

1 IntroductionPreliminary InformationContribution to the ThesisTime Plan

2 Algorithm Design

3 Evaluation

4 Results

5 Conclusion

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 35: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Publications

Aral, A. and Ovatman, T. (2014). Improving resource utilization in cloudenvironments using application placement heuristics. In Proceedings of the4th International Conference on Cloud Computing and Services Science(CLOSER), pages 527–534.Aral, A. and Ovatman, T. (2015). Subgraph matching for resource allocation inthe federated cloud environment. In Proceedings of 8th IEEE InternationalConference on Cloud Computing (IEEE CLOUD). (to appear)Aral, A. and Ovatman, T. (2015). Graph theoretical allocation of map reduceclusters in federated cloud. (for journal submission)

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 36: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Planned Studies

AlgorithmAdditional constraints (jurisdiction, partially known topology)Vertical scaling supportHybrid cloud supportHomeomorphismConnected components

EvaluationSignificance studyEvaluation with topology improvementsMulti-objective optimizationDynamic heuristic selection, meta-heuristics

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud

Page 37: Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progress 2]

IntroductionAlgorithm Design

EvaluationResults

Conclusion

Thank you for your time.

Atakan Aral Modeling and Optimization of Resource Allocation in Cloud