SlideShare a Scribd company logo
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
DPSO
An Optimization Approach for Load Balancing in Parallel
Link Discovery
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo
September 17, 2015
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 1/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Outline
1 Motivation
2 Load Balancing Approaches
3 Evaluation
4 Conclusion and Future Work
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 2/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Outline
1 Motivation
2 Load Balancing Approaches
3 Evaluation
4 Conclusion and Future Work
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 3/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Link Discovery (LD)
Why LD?
1 Fourth principle
2 Links are central for
Cross-ontology QA
Data Integration
Reasoning
Federated Queries
...
LD Time complexity
Large number of triples (> 63 billion triples)
Quadratic runtime
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 4/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Load Balancing for LD
Need to develop highly scalable LD algorithms
Local hardware LD
Suffer less from the data transfer bottleneck
Better runtime than parallel LD approaches on remote
hardware (e.g. cloud-based approaches)
Current load balancing approaches for local LD
Paid little attention
Mostly na¨ıve implementations
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 5/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Load Balancing for LD
Need to develop highly scalable LD algorithms
Local hardware LD
Suffer less from the data transfer bottleneck
Better runtime than parallel LD approaches on remote
hardware (e.g. cloud-based approaches)
Current load balancing approaches for local LD
Paid little attention
Mostly na¨ıve implementations
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 5/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Load Balancing for LD
Need to develop highly scalable LD algorithms
Local hardware LD
Suffer less from the data transfer bottleneck
Better runtime than parallel LD approaches on remote
hardware (e.g. cloud-based approaches)
Current load balancing approaches for local LD
Paid little attention
Mostly na¨ıve implementations
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 5/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Load Balancing Problem
Given:
n tasks τ1, ..., τn
Computational complexities c(τ1), ..., c(τn)
m processors
Goal:
Distribute τi across m processors as evenly as possible
Example
3 tasks τ1, τ2 and τ3 with complexities 3, 4 resp. 6
2 processors
An optimal distribution:
P1 → {τ1, τ2}
P2 → {τ3}
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 6/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Load Balancing Problem
Given:
n tasks τ1, ..., τn
Computational complexities c(τ1), ..., c(τn)
m processors
Goal:
Distribute τi across m processors as evenly as possible
Example
3 tasks τ1, τ2 and τ3 with complexities 3, 4 resp. 6
2 processors
An optimal distribution:
P1 → {τ1, τ2}
P2 → {τ3}
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 6/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Outline
1 Motivation
2 Load Balancing Approaches
3 Evaluation
4 Conclusion and Future Work
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 7/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Load Balancing for LD
Notations
Given S, T of source resp. target resources
Divides S, T such that
k
i=1
Si = S and
l
j=1
Tj = T
Determines (Si , Tj ) whose elements are to be compared
The idea of load balancing is to distribute the
computation of Si × Tj over m processors
Task τ: Comparing elements in (Si , Tj )
c(τ) = |Si | · |Tj |
block B: Set of all tasks assigned to a single processor
c(B) =
t∈B
c(τ)
MSE = m
i=1 c(Bi ) − m
j=1
c(Bj )
m
2
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Load Balancing for LD
Notations
Given S, T of source resp. target resources
Divides S, T such that
k
i=1
Si = S and
l
j=1
Tj = T
Determines (Si , Tj ) whose elements are to be compared
The idea of load balancing is to distribute the
computation of Si × Tj over m processors
Task τ: Comparing elements in (Si , Tj )
c(τ) = |Si | · |Tj |
block B: Set of all tasks assigned to a single processor
c(B) =
t∈B
c(τ)
MSE = m
i=1 c(Bi ) − m
j=1
c(Bj )
m
2
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Load Balancing for LD
Notations
Given S, T of source resp. target resources
Divides S, T such that
k
i=1
Si = S and
l
j=1
Tj = T
Determines (Si , Tj ) whose elements are to be compared
The idea of load balancing is to distribute the
computation of Si × Tj over m processors
Task τ: Comparing elements in (Si , Tj )
c(τ) = |Si | · |Tj |
block B: Set of all tasks assigned to a single processor
c(B) =
t∈B
c(τ)
MSE = m
i=1 c(Bi ) − m
j=1
c(Bj )
m
2
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Load Balancing for LD
Notations
Given S, T of source resp. target resources
Divides S, T such that
k
i=1
Si = S and
l
j=1
Tj = T
Determines (Si , Tj ) whose elements are to be compared
The idea of load balancing is to distribute the
computation of Si × Tj over m processors
Task τ: Comparing elements in (Si , Tj )
c(τ) = |Si | · |Tj |
block B: Set of all tasks assigned to a single processor
c(B) =
t∈B
c(τ)
MSE = m
i=1 c(Bi ) − m
j=1
c(Bj )
m
2
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Load Balancing for LD
Notations
Given S, T of source resp. target resources
Divides S, T such that
k
i=1
Si = S and
l
j=1
Tj = T
Determines (Si , Tj ) whose elements are to be compared
The idea of load balancing is to distribute the
computation of Si × Tj over m processors
Task τ: Comparing elements in (Si , Tj )
c(τ) = |Si | · |Tj |
block B: Set of all tasks assigned to a single processor
c(B) =
t∈B
c(τ)
MSE = m
i=1 c(Bi ) − m
j=1
c(Bj )
m
2
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Load Balancing for LD
Notations
Given S, T of source resp. target resources
Divides S, T such that
k
i=1
Si = S and
l
j=1
Tj = T
Determines (Si , Tj ) whose elements are to be compared
The idea of load balancing is to distribute the
computation of Si × Tj over m processors
Task τ: Comparing elements in (Si , Tj )
c(τ) = |Si | · |Tj |
block B: Set of all tasks assigned to a single processor
c(B) =
t∈B
c(τ)
MSE = m
i=1 c(Bi ) − m
j=1
c(Bj )
m
2
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Load Balancing for LD
Notations
Given S, T of source resp. target resources
Divides S, T such that
k
i=1
Si = S and
l
j=1
Tj = T
Determines (Si , Tj ) whose elements are to be compared
The idea of load balancing is to distribute the
computation of Si × Tj over m processors
Task τ: Comparing elements in (Si , Tj )
c(τ) = |Si | · |Tj |
block B: Set of all tasks assigned to a single processor
c(B) =
t∈B
c(τ)
MSE = m
i=1 c(Bi ) − m
j=1
c(Bj )
m
2
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Running Example
4 tasks {τ1, τ2, τ3, τ4}
Respective complexities {7, 1, 8, 3}
2 processors P1, P2
Tasks: 7 1 8 3
Processors: 1 1
τ1 assigned to P1: 7
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 9/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Na¨ıve Load Balancer
Idea
Divides tasks between processors based on their index and
regardless of complexity
Example
Processors assignment: 7 1 8 3 MSE = 30.25
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 10/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Greedy Load Balancer
Idea
1 Sorts tasks in descending order based on their complexity
2 Starting from the most complex task, assigns tasks to
processors in order
Example
1. Sorted tasks: 8 7 3 1
2. Processors assignment: 8 7 3 1
MSE = 2.25
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 11/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Pair-Based Load Balancer
Idea
1 Sorts tasks according to task complexity
2 In order, assigns ith
and (n − i + 1)th
tasks to Pi
Example
1. Sort tasks: 1 3 7 8
2. Processors assignment: 1 3 7 8
MSE = 0.25
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 12/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Particle Swarm Optimization
Idea
Initialization
Best Known Positions (BKP)
BKP ← Partitions the n tasks to m task blocks
Computes fitness function F to the initial BKP
F is the complexity difference between the most and
least loaded blocks
Initializes Best Known Fitness (BKF) to F
Until a termination criterion is met
Performs the particles migration, based on random
particle velocity
Recomputes F
If F < BKF, updates BKF and BKP
Return BKP
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 13/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Particle Swarm Optimization
Idea
Initialization
Best Known Positions (BKP)
BKP ← Partitions the n tasks to m task blocks
Computes fitness function F to the initial BKP
F is the complexity difference between the most and
least loaded blocks
Initializes Best Known Fitness (BKF) to F
Until a termination criterion is met
Performs the particles migration, based on random
particle velocity
Recomputes F
If F < BKF, updates BKF and BKP
Return BKP
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 13/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Particle Swarm Optimization
Idea
Initialization
Best Known Positions (BKP)
BKP ← Partitions the n tasks to m task blocks
Computes fitness function F to the initial BKP
F is the complexity difference between the most and
least loaded blocks
Initializes Best Known Fitness (BKF) to F
Until a termination criterion is met
Performs the particles migration, based on random
particle velocity
Recomputes F
If F < BKF, updates BKF and BKP
Return BKP
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 13/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Particle Swarm Optimization
Idea
Initialization
Best Known Positions (BKP)
BKP ← Partitions the n tasks to m task blocks
Computes fitness function F to the initial BKP
F is the complexity difference between the most and
least loaded blocks
Initializes Best Known Fitness (BKF) to F
Until a termination criterion is met
Performs the particles migration, based on random
particle velocity
Recomputes F
If F < BKF, updates BKF and BKP
Return BKP
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 13/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Particle Swarm Optimization
Example
Termination criterion: max number of iterations of 1
Initialization:
7 1 8 3 BKF = F = 11
First iteration:
7 1 8 3 F = 5
As F < BKF, updates BKF and PKB
MSE = 6.25
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 14/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Deterministic PSO (DPSO)
DPSO
PSO is non-deterministic
PSO depends on a random selection of velocity for
moving particles
We propose the Deterministic PSO (DPSO)
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 15/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Deterministic PSO (DPSO)
Idea
Partitions the n tasks to m task blocks
Until termination criterion is met
Finds the most overloaded block B+ and the least
underloaded block B−
Sorts tasks within B+ based in their complexities
As far as a better balancing between B+ to B− is met
Moves tasks in order from B+
to B−
(task migration)
Computes fitness function as complexity difference
between B+ and B−
Returns best known blocks
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 16/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Deterministic PSO (DPSO)
Why is DPSO deterministic?
Only moves tasks from B+
to B−
(no random migration)
Sorts B+
tasks before task migration start
Insures optimal load balancing between B+
and B−
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 17/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Deterministic PSO (DPSO)
Example
Termination criterion: max number of iterations of 1
Initialization:
B+
= 8 7 , B−
= 1 3 F = 11
First iteration:
Sorted B+
= 7 8 , B−
= 1 3
Task migration: B+
= 8 , B−
= 1 3 7 F = 3
MSE = 2.25
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 18/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Outline
1 Motivation
2 Load Balancing Approaches
3 Evaluation
4 Conclusion and Future Work
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 19/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Evaluation Setup
Orchid
The parallel task generation was based on Orchid
Orchid partitions the surface of the planet
A task is the comparison of all points in two partitions
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 20/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Evaluation Setup
Datasets
1 5 synthetic geographic datasets
Polygons’ sizes varied from 1 to 10 points
2 3 real geographic datasets
Nuts
DBpedia
LinkedGeoData
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 21/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Evaluation Setup
Hardware
64-core server running OpenJDK 64-Bit Server 1.6.0 27
on Ubuntu 12.04.2 LTS.
8 quad-core processor Intel(R) Core(TM) i7-3770 CPU @
3.40 GHz with 8192 KB cache
Each experiment was assigned 20 GB of memory
PSO
PSO ran 5 times in each experiment and provide the
mean of the 5 runs
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 22/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Orchid vs. Parallel Orchid
Goal
Evaluate the speedup gained by using parallel Orchid
For Nuts, PSO and DPSO up to 3 times faster
For LinkedGeoData, PSO and DPSO up to 10 times faster
2 4 8
Number of threads
0.0
0.1
0.2
0.3
0.4
Time(min.)
Naïve
Greedy
PairBased
PSO
DPSO
Orchid
Nuts runtime
2 4 8
Number of threads
1
10
100
1000
Time(min.)
Naïve
Greedy
PairBased
PSO
DPSO
Orchid
LinkedGeoData runtime
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 23/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Orchid vs. Parallel Orchid
Goal
PSO and DPSO are capable of achieving superlinear
performance, as processor caches are faster than RAM
Greedy and pair-based fail to achieve even the run time of
the normal Orchid, due to significant sorting time
2 4 8
Number of threads
0.0
0.1
0.2
0.3
0.4
Time(min.)
Naïve
Greedy
PairBased
PSO
DPSO
Orchid
Nuts runtime
2 4 8
Number of threads
1
10
100
1000
Time(min.)
Naïve
Greedy
PairBased
PSO
DPSO
Orchid
LinkedGeoData runtime
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 24/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Parallel Load balancing Algorithms
Goals
Measure each algorithm runtime
Qualify each algorithm data distribution using MSE
2 4 8
Number of threads
0.0
0.1
0.2
0.3
0.4
Time(min.)
Naïve
Greedy
PairBased
PSO
DPSO
Orchid
Nuts runtime
2 4 8
Number of threads
1010
1011
1012
1013
MSE
Naïve
Greedy
PairBased
PSO
DPSO
Nuts MSE
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 25/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Parallel Load balancing Algorithms
Goals
Measure each algorithm runtime
Qualify each algorithm data distribution using MSE
2 4 8
Number of threads
0.0
0.1
1.0
10.0
Time(min.)
Naïve
Greedy
PairBased
PSO
DPSO
Orchid
DBpedia runtime
2 4 8
Number of threads
1011
1012
1013
MSE
Naïve
Greedy
PairBased
PSO
DPSO
DBpedia MSE
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 26/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Parallel Load balancing Algorithms
Goals
Measure each algorithm runtime
Qualify each algorithm data distribution using MSE
2 4 8
Number of threads
1
10
100
1000
Time(min.)
Naïve
Greedy
PairBased
PSO
DPSO
Orchid
LinkedGeoData runtime
2 4 8
Number of threads
1009
1010
1011
MSE
Naïve
Greedy
PairBased
PSO
DPSO
LinkedGeoData MSE
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 27/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Outline
1 Motivation
2 Load Balancing Approaches
3 Evaluation
4 Conclusion and Future Work
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 28/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Conclusion and Future Work
Conclusion
Presented load balancing techniques for link discovery
Proposed deterministic PSO (DPSO)
Combined load balancing algorithms with Orchid
Evaluated on real and artificial datasets
Future Work
Enable splitting of one task over multiple processors
Implement a caching techniques
Study the combination of DPSO with other task
generation algorithms
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 29/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Conclusion and Future Work
Conclusion
Presented load balancing techniques for link discovery
Proposed deterministic PSO (DPSO)
Combined load balancing algorithms with Orchid
Evaluated on real and artificial datasets
Future Work
Enable splitting of one task over multiple processors
Implement a caching techniques
Study the combination of DPSO with other task
generation algorithms
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 29/30
Motivation Load Balancing Approaches Evaluation Conclusion and Future Work
Thank You!
Questions?
Mohamed Sherif
Augustusplatz 10
D-04109 Leipzig
sherif@informatik.uni-leipzig.de
http://aksw.org/MohamedSherif
http://aksw.org/Projects/LIMES
#akswgroup
Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 30/30

More Related Content

Similar to Dpso -- An Optimization Approach for Load Balancing in Parallel Link Discovery

Duality Theory in Multi Objective Linear Programming Problems
Duality Theory in Multi Objective Linear Programming ProblemsDuality Theory in Multi Objective Linear Programming Problems
Duality Theory in Multi Objective Linear Programming Problems
theijes
 
Size Measurement and Estimation
Size Measurement and EstimationSize Measurement and Estimation
Size Measurement and Estimation
Louis A. Poulin
 
Comparisional Investigation of Load Dispatch Solutions with TLBO
Comparisional Investigation of Load Dispatch Solutions with TLBO Comparisional Investigation of Load Dispatch Solutions with TLBO
Comparisional Investigation of Load Dispatch Solutions with TLBO
IJECEIAES
 
An efficient method of solving lexicographic linear goal programming problem
An efficient method of solving lexicographic linear goal programming problemAn efficient method of solving lexicographic linear goal programming problem
An efficient method of solving lexicographic linear goal programming problemAlexander Decker
 
An efficient method of solving lexicographic linear goal programming problem
An efficient method of solving lexicographic linear goal programming problemAn efficient method of solving lexicographic linear goal programming problem
An efficient method of solving lexicographic linear goal programming problemAlexander Decker
 
Chapter 5 - PERT-CPM
Chapter 5 - PERT-CPM Chapter 5 - PERT-CPM
Chapter 5 - PERT-CPM
maggieget
 
Size measurement and estimation
Size measurement and estimationSize measurement and estimation
Size measurement and estimation
Louis A. Poulin
 
A Heterogeneous Static Hierarchical Expected Completion Time Based Scheduling...
A Heterogeneous Static Hierarchical Expected Completion Time Based Scheduling...A Heterogeneous Static Hierarchical Expected Completion Time Based Scheduling...
A Heterogeneous Static Hierarchical Expected Completion Time Based Scheduling...
IRJET Journal
 
A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...
A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...
A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...Cemal Ardil
 
Particle Swarm Optimization in the fine-tuning of Fuzzy Software Cost Estimat...
Particle Swarm Optimization in the fine-tuning of Fuzzy Software Cost Estimat...Particle Swarm Optimization in the fine-tuning of Fuzzy Software Cost Estimat...
Particle Swarm Optimization in the fine-tuning of Fuzzy Software Cost Estimat...
Waqas Tariq
 
Assingment Problem3
Assingment Problem3Assingment Problem3
Assingment Problem3Evren E
 
Automatically Estimating Software Effort and Cost using Computing Intelligenc...
Automatically Estimating Software Effort and Cost using Computing Intelligenc...Automatically Estimating Software Effort and Cost using Computing Intelligenc...
Automatically Estimating Software Effort and Cost using Computing Intelligenc...
cscpconf
 
An Effective PSO-inspired Algorithm for Workflow Scheduling
An Effective PSO-inspired Algorithm for Workflow Scheduling An Effective PSO-inspired Algorithm for Workflow Scheduling
An Effective PSO-inspired Algorithm for Workflow Scheduling
IJECEIAES
 
Software Effort Estimation Using Particle Swarm Optimization with Inertia Weight
Software Effort Estimation Using Particle Swarm Optimization with Inertia WeightSoftware Effort Estimation Using Particle Swarm Optimization with Inertia Weight
Software Effort Estimation Using Particle Swarm Optimization with Inertia Weight
Waqas Tariq
 
Ga
GaGa
Boosting auxiliary task guidance: a probabilistic approach
Boosting auxiliary task guidance: a probabilistic approachBoosting auxiliary task guidance: a probabilistic approach
Boosting auxiliary task guidance: a probabilistic approach
IAESIJAI
 
Advanced SOM & K Mean Method for Load Curve Clustering
Advanced SOM & K Mean Method for Load Curve Clustering Advanced SOM & K Mean Method for Load Curve Clustering
Advanced SOM & K Mean Method for Load Curve Clustering
IJECEIAES
 
Resource management
Resource managementResource management
Resource management
Dr Sandeep Kumar Poonia
 

Similar to Dpso -- An Optimization Approach for Load Balancing in Parallel Link Discovery (20)

Duality Theory in Multi Objective Linear Programming Problems
Duality Theory in Multi Objective Linear Programming ProblemsDuality Theory in Multi Objective Linear Programming Problems
Duality Theory in Multi Objective Linear Programming Problems
 
Size Measurement and Estimation
Size Measurement and EstimationSize Measurement and Estimation
Size Measurement and Estimation
 
Comparisional Investigation of Load Dispatch Solutions with TLBO
Comparisional Investigation of Load Dispatch Solutions with TLBO Comparisional Investigation of Load Dispatch Solutions with TLBO
Comparisional Investigation of Load Dispatch Solutions with TLBO
 
Project Management Techniques
Project Management TechniquesProject Management Techniques
Project Management Techniques
 
An efficient method of solving lexicographic linear goal programming problem
An efficient method of solving lexicographic linear goal programming problemAn efficient method of solving lexicographic linear goal programming problem
An efficient method of solving lexicographic linear goal programming problem
 
An efficient method of solving lexicographic linear goal programming problem
An efficient method of solving lexicographic linear goal programming problemAn efficient method of solving lexicographic linear goal programming problem
An efficient method of solving lexicographic linear goal programming problem
 
Chapter 5 - PERT-CPM
Chapter 5 - PERT-CPM Chapter 5 - PERT-CPM
Chapter 5 - PERT-CPM
 
Size measurement and estimation
Size measurement and estimationSize measurement and estimation
Size measurement and estimation
 
A Heterogeneous Static Hierarchical Expected Completion Time Based Scheduling...
A Heterogeneous Static Hierarchical Expected Completion Time Based Scheduling...A Heterogeneous Static Hierarchical Expected Completion Time Based Scheduling...
A Heterogeneous Static Hierarchical Expected Completion Time Based Scheduling...
 
A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...
A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...
A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...
 
Particle Swarm Optimization in the fine-tuning of Fuzzy Software Cost Estimat...
Particle Swarm Optimization in the fine-tuning of Fuzzy Software Cost Estimat...Particle Swarm Optimization in the fine-tuning of Fuzzy Software Cost Estimat...
Particle Swarm Optimization in the fine-tuning of Fuzzy Software Cost Estimat...
 
Assingment Problem3
Assingment Problem3Assingment Problem3
Assingment Problem3
 
Ga
GaGa
Ga
 
Automatically Estimating Software Effort and Cost using Computing Intelligenc...
Automatically Estimating Software Effort and Cost using Computing Intelligenc...Automatically Estimating Software Effort and Cost using Computing Intelligenc...
Automatically Estimating Software Effort and Cost using Computing Intelligenc...
 
An Effective PSO-inspired Algorithm for Workflow Scheduling
An Effective PSO-inspired Algorithm for Workflow Scheduling An Effective PSO-inspired Algorithm for Workflow Scheduling
An Effective PSO-inspired Algorithm for Workflow Scheduling
 
Software Effort Estimation Using Particle Swarm Optimization with Inertia Weight
Software Effort Estimation Using Particle Swarm Optimization with Inertia WeightSoftware Effort Estimation Using Particle Swarm Optimization with Inertia Weight
Software Effort Estimation Using Particle Swarm Optimization with Inertia Weight
 
Ga
GaGa
Ga
 
Boosting auxiliary task guidance: a probabilistic approach
Boosting auxiliary task guidance: a probabilistic approachBoosting auxiliary task guidance: a probabilistic approach
Boosting auxiliary task guidance: a probabilistic approach
 
Advanced SOM & K Mean Method for Load Curve Clustering
Advanced SOM & K Mean Method for Load Curve Clustering Advanced SOM & K Mean Method for Load Curve Clustering
Advanced SOM & K Mean Method for Load Curve Clustering
 
Resource management
Resource managementResource management
Resource management
 

Recently uploaded

Acetabularia Information For Class 9 .docx
Acetabularia Information For Class 9  .docxAcetabularia Information For Class 9  .docx
Acetabularia Information For Class 9 .docx
vaibhavrinwa19
 
June 3, 2024 Anti-Semitism Letter Sent to MIT President Kornbluth and MIT Cor...
June 3, 2024 Anti-Semitism Letter Sent to MIT President Kornbluth and MIT Cor...June 3, 2024 Anti-Semitism Letter Sent to MIT President Kornbluth and MIT Cor...
June 3, 2024 Anti-Semitism Letter Sent to MIT President Kornbluth and MIT Cor...
Levi Shapiro
 
Executive Directors Chat Leveraging AI for Diversity, Equity, and Inclusion
Executive Directors Chat  Leveraging AI for Diversity, Equity, and InclusionExecutive Directors Chat  Leveraging AI for Diversity, Equity, and Inclusion
Executive Directors Chat Leveraging AI for Diversity, Equity, and Inclusion
TechSoup
 
MASS MEDIA STUDIES-835-CLASS XI Resource Material.pdf
MASS MEDIA STUDIES-835-CLASS XI Resource Material.pdfMASS MEDIA STUDIES-835-CLASS XI Resource Material.pdf
MASS MEDIA STUDIES-835-CLASS XI Resource Material.pdf
goswamiyash170123
 
South African Journal of Science: Writing with integrity workshop (2024)
South African Journal of Science: Writing with integrity workshop (2024)South African Journal of Science: Writing with integrity workshop (2024)
South African Journal of Science: Writing with integrity workshop (2024)
Academy of Science of South Africa
 
Unit 8 - Information and Communication Technology (Paper I).pdf
Unit 8 - Information and Communication Technology (Paper I).pdfUnit 8 - Information and Communication Technology (Paper I).pdf
Unit 8 - Information and Communication Technology (Paper I).pdf
Thiyagu K
 
A Strategic Approach: GenAI in Education
A Strategic Approach: GenAI in EducationA Strategic Approach: GenAI in Education
A Strategic Approach: GenAI in Education
Peter Windle
 
S1-Introduction-Biopesticides in ICM.pptx
S1-Introduction-Biopesticides in ICM.pptxS1-Introduction-Biopesticides in ICM.pptx
S1-Introduction-Biopesticides in ICM.pptx
tarandeep35
 
Natural birth techniques - Mrs.Akanksha Trivedi Rama University
Natural birth techniques - Mrs.Akanksha Trivedi Rama UniversityNatural birth techniques - Mrs.Akanksha Trivedi Rama University
Natural birth techniques - Mrs.Akanksha Trivedi Rama University
Akanksha trivedi rama nursing college kanpur.
 
Lapbook sobre os Regimes Totalitários.pdf
Lapbook sobre os Regimes Totalitários.pdfLapbook sobre os Regimes Totalitários.pdf
Lapbook sobre os Regimes Totalitários.pdf
Jean Carlos Nunes Paixão
 
Synthetic Fiber Construction in lab .pptx
Synthetic Fiber Construction in lab .pptxSynthetic Fiber Construction in lab .pptx
Synthetic Fiber Construction in lab .pptx
Pavel ( NSTU)
 
Chapter 3 - Islamic Banking Products and Services.pptx
Chapter 3 - Islamic Banking Products and Services.pptxChapter 3 - Islamic Banking Products and Services.pptx
Chapter 3 - Islamic Banking Products and Services.pptx
Mohd Adib Abd Muin, Senior Lecturer at Universiti Utara Malaysia
 
Model Attribute Check Company Auto Property
Model Attribute  Check Company Auto PropertyModel Attribute  Check Company Auto Property
Model Attribute Check Company Auto Property
Celine George
 
How libraries can support authors with open access requirements for UKRI fund...
How libraries can support authors with open access requirements for UKRI fund...How libraries can support authors with open access requirements for UKRI fund...
How libraries can support authors with open access requirements for UKRI fund...
Jisc
 
A Survey of Techniques for Maximizing LLM Performance.pptx
A Survey of Techniques for Maximizing LLM Performance.pptxA Survey of Techniques for Maximizing LLM Performance.pptx
A Survey of Techniques for Maximizing LLM Performance.pptx
thanhdowork
 
Thesis Statement for students diagnonsed withADHD.ppt
Thesis Statement for students diagnonsed withADHD.pptThesis Statement for students diagnonsed withADHD.ppt
Thesis Statement for students diagnonsed withADHD.ppt
EverAndrsGuerraGuerr
 
BÀI TẬP BỔ TRỢ TIẾNG ANH GLOBAL SUCCESS LỚP 3 - CẢ NĂM (CÓ FILE NGHE VÀ ĐÁP Á...
BÀI TẬP BỔ TRỢ TIẾNG ANH GLOBAL SUCCESS LỚP 3 - CẢ NĂM (CÓ FILE NGHE VÀ ĐÁP Á...BÀI TẬP BỔ TRỢ TIẾNG ANH GLOBAL SUCCESS LỚP 3 - CẢ NĂM (CÓ FILE NGHE VÀ ĐÁP Á...
BÀI TẬP BỔ TRỢ TIẾNG ANH GLOBAL SUCCESS LỚP 3 - CẢ NĂM (CÓ FILE NGHE VÀ ĐÁP Á...
Nguyen Thanh Tu Collection
 
"Protectable subject matters, Protection in biotechnology, Protection of othe...
"Protectable subject matters, Protection in biotechnology, Protection of othe..."Protectable subject matters, Protection in biotechnology, Protection of othe...
"Protectable subject matters, Protection in biotechnology, Protection of othe...
SACHIN R KONDAGURI
 
JEE1_This_section_contains_FOUR_ questions
JEE1_This_section_contains_FOUR_ questionsJEE1_This_section_contains_FOUR_ questions
JEE1_This_section_contains_FOUR_ questions
ShivajiThube2
 
Digital Artifact 2 - Investigating Pavilion Designs
Digital Artifact 2 - Investigating Pavilion DesignsDigital Artifact 2 - Investigating Pavilion Designs
Digital Artifact 2 - Investigating Pavilion Designs
chanes7
 

Recently uploaded (20)

Acetabularia Information For Class 9 .docx
Acetabularia Information For Class 9  .docxAcetabularia Information For Class 9  .docx
Acetabularia Information For Class 9 .docx
 
June 3, 2024 Anti-Semitism Letter Sent to MIT President Kornbluth and MIT Cor...
June 3, 2024 Anti-Semitism Letter Sent to MIT President Kornbluth and MIT Cor...June 3, 2024 Anti-Semitism Letter Sent to MIT President Kornbluth and MIT Cor...
June 3, 2024 Anti-Semitism Letter Sent to MIT President Kornbluth and MIT Cor...
 
Executive Directors Chat Leveraging AI for Diversity, Equity, and Inclusion
Executive Directors Chat  Leveraging AI for Diversity, Equity, and InclusionExecutive Directors Chat  Leveraging AI for Diversity, Equity, and Inclusion
Executive Directors Chat Leveraging AI for Diversity, Equity, and Inclusion
 
MASS MEDIA STUDIES-835-CLASS XI Resource Material.pdf
MASS MEDIA STUDIES-835-CLASS XI Resource Material.pdfMASS MEDIA STUDIES-835-CLASS XI Resource Material.pdf
MASS MEDIA STUDIES-835-CLASS XI Resource Material.pdf
 
South African Journal of Science: Writing with integrity workshop (2024)
South African Journal of Science: Writing with integrity workshop (2024)South African Journal of Science: Writing with integrity workshop (2024)
South African Journal of Science: Writing with integrity workshop (2024)
 
Unit 8 - Information and Communication Technology (Paper I).pdf
Unit 8 - Information and Communication Technology (Paper I).pdfUnit 8 - Information and Communication Technology (Paper I).pdf
Unit 8 - Information and Communication Technology (Paper I).pdf
 
A Strategic Approach: GenAI in Education
A Strategic Approach: GenAI in EducationA Strategic Approach: GenAI in Education
A Strategic Approach: GenAI in Education
 
S1-Introduction-Biopesticides in ICM.pptx
S1-Introduction-Biopesticides in ICM.pptxS1-Introduction-Biopesticides in ICM.pptx
S1-Introduction-Biopesticides in ICM.pptx
 
Natural birth techniques - Mrs.Akanksha Trivedi Rama University
Natural birth techniques - Mrs.Akanksha Trivedi Rama UniversityNatural birth techniques - Mrs.Akanksha Trivedi Rama University
Natural birth techniques - Mrs.Akanksha Trivedi Rama University
 
Lapbook sobre os Regimes Totalitários.pdf
Lapbook sobre os Regimes Totalitários.pdfLapbook sobre os Regimes Totalitários.pdf
Lapbook sobre os Regimes Totalitários.pdf
 
Synthetic Fiber Construction in lab .pptx
Synthetic Fiber Construction in lab .pptxSynthetic Fiber Construction in lab .pptx
Synthetic Fiber Construction in lab .pptx
 
Chapter 3 - Islamic Banking Products and Services.pptx
Chapter 3 - Islamic Banking Products and Services.pptxChapter 3 - Islamic Banking Products and Services.pptx
Chapter 3 - Islamic Banking Products and Services.pptx
 
Model Attribute Check Company Auto Property
Model Attribute  Check Company Auto PropertyModel Attribute  Check Company Auto Property
Model Attribute Check Company Auto Property
 
How libraries can support authors with open access requirements for UKRI fund...
How libraries can support authors with open access requirements for UKRI fund...How libraries can support authors with open access requirements for UKRI fund...
How libraries can support authors with open access requirements for UKRI fund...
 
A Survey of Techniques for Maximizing LLM Performance.pptx
A Survey of Techniques for Maximizing LLM Performance.pptxA Survey of Techniques for Maximizing LLM Performance.pptx
A Survey of Techniques for Maximizing LLM Performance.pptx
 
Thesis Statement for students diagnonsed withADHD.ppt
Thesis Statement for students diagnonsed withADHD.pptThesis Statement for students diagnonsed withADHD.ppt
Thesis Statement for students diagnonsed withADHD.ppt
 
BÀI TẬP BỔ TRỢ TIẾNG ANH GLOBAL SUCCESS LỚP 3 - CẢ NĂM (CÓ FILE NGHE VÀ ĐÁP Á...
BÀI TẬP BỔ TRỢ TIẾNG ANH GLOBAL SUCCESS LỚP 3 - CẢ NĂM (CÓ FILE NGHE VÀ ĐÁP Á...BÀI TẬP BỔ TRỢ TIẾNG ANH GLOBAL SUCCESS LỚP 3 - CẢ NĂM (CÓ FILE NGHE VÀ ĐÁP Á...
BÀI TẬP BỔ TRỢ TIẾNG ANH GLOBAL SUCCESS LỚP 3 - CẢ NĂM (CÓ FILE NGHE VÀ ĐÁP Á...
 
"Protectable subject matters, Protection in biotechnology, Protection of othe...
"Protectable subject matters, Protection in biotechnology, Protection of othe..."Protectable subject matters, Protection in biotechnology, Protection of othe...
"Protectable subject matters, Protection in biotechnology, Protection of othe...
 
JEE1_This_section_contains_FOUR_ questions
JEE1_This_section_contains_FOUR_ questionsJEE1_This_section_contains_FOUR_ questions
JEE1_This_section_contains_FOUR_ questions
 
Digital Artifact 2 - Investigating Pavilion Designs
Digital Artifact 2 - Investigating Pavilion DesignsDigital Artifact 2 - Investigating Pavilion Designs
Digital Artifact 2 - Investigating Pavilion Designs
 

Dpso -- An Optimization Approach for Load Balancing in Parallel Link Discovery

  • 1. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work DPSO An Optimization Approach for Load Balancing in Parallel Link Discovery Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo September 17, 2015 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 1/30
  • 2. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Outline 1 Motivation 2 Load Balancing Approaches 3 Evaluation 4 Conclusion and Future Work Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 2/30
  • 3. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Outline 1 Motivation 2 Load Balancing Approaches 3 Evaluation 4 Conclusion and Future Work Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 3/30
  • 4. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Link Discovery (LD) Why LD? 1 Fourth principle 2 Links are central for Cross-ontology QA Data Integration Reasoning Federated Queries ... LD Time complexity Large number of triples (> 63 billion triples) Quadratic runtime Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 4/30
  • 5. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Load Balancing for LD Need to develop highly scalable LD algorithms Local hardware LD Suffer less from the data transfer bottleneck Better runtime than parallel LD approaches on remote hardware (e.g. cloud-based approaches) Current load balancing approaches for local LD Paid little attention Mostly na¨ıve implementations Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 5/30
  • 6. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Load Balancing for LD Need to develop highly scalable LD algorithms Local hardware LD Suffer less from the data transfer bottleneck Better runtime than parallel LD approaches on remote hardware (e.g. cloud-based approaches) Current load balancing approaches for local LD Paid little attention Mostly na¨ıve implementations Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 5/30
  • 7. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Load Balancing for LD Need to develop highly scalable LD algorithms Local hardware LD Suffer less from the data transfer bottleneck Better runtime than parallel LD approaches on remote hardware (e.g. cloud-based approaches) Current load balancing approaches for local LD Paid little attention Mostly na¨ıve implementations Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 5/30
  • 8. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Load Balancing Problem Given: n tasks τ1, ..., τn Computational complexities c(τ1), ..., c(τn) m processors Goal: Distribute τi across m processors as evenly as possible Example 3 tasks τ1, τ2 and τ3 with complexities 3, 4 resp. 6 2 processors An optimal distribution: P1 → {τ1, τ2} P2 → {τ3} Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 6/30
  • 9. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Load Balancing Problem Given: n tasks τ1, ..., τn Computational complexities c(τ1), ..., c(τn) m processors Goal: Distribute τi across m processors as evenly as possible Example 3 tasks τ1, τ2 and τ3 with complexities 3, 4 resp. 6 2 processors An optimal distribution: P1 → {τ1, τ2} P2 → {τ3} Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 6/30
  • 10. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Outline 1 Motivation 2 Load Balancing Approaches 3 Evaluation 4 Conclusion and Future Work Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 7/30
  • 11. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Load Balancing for LD Notations Given S, T of source resp. target resources Divides S, T such that k i=1 Si = S and l j=1 Tj = T Determines (Si , Tj ) whose elements are to be compared The idea of load balancing is to distribute the computation of Si × Tj over m processors Task τ: Comparing elements in (Si , Tj ) c(τ) = |Si | · |Tj | block B: Set of all tasks assigned to a single processor c(B) = t∈B c(τ) MSE = m i=1 c(Bi ) − m j=1 c(Bj ) m 2 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
  • 12. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Load Balancing for LD Notations Given S, T of source resp. target resources Divides S, T such that k i=1 Si = S and l j=1 Tj = T Determines (Si , Tj ) whose elements are to be compared The idea of load balancing is to distribute the computation of Si × Tj over m processors Task τ: Comparing elements in (Si , Tj ) c(τ) = |Si | · |Tj | block B: Set of all tasks assigned to a single processor c(B) = t∈B c(τ) MSE = m i=1 c(Bi ) − m j=1 c(Bj ) m 2 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
  • 13. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Load Balancing for LD Notations Given S, T of source resp. target resources Divides S, T such that k i=1 Si = S and l j=1 Tj = T Determines (Si , Tj ) whose elements are to be compared The idea of load balancing is to distribute the computation of Si × Tj over m processors Task τ: Comparing elements in (Si , Tj ) c(τ) = |Si | · |Tj | block B: Set of all tasks assigned to a single processor c(B) = t∈B c(τ) MSE = m i=1 c(Bi ) − m j=1 c(Bj ) m 2 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
  • 14. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Load Balancing for LD Notations Given S, T of source resp. target resources Divides S, T such that k i=1 Si = S and l j=1 Tj = T Determines (Si , Tj ) whose elements are to be compared The idea of load balancing is to distribute the computation of Si × Tj over m processors Task τ: Comparing elements in (Si , Tj ) c(τ) = |Si | · |Tj | block B: Set of all tasks assigned to a single processor c(B) = t∈B c(τ) MSE = m i=1 c(Bi ) − m j=1 c(Bj ) m 2 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
  • 15. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Load Balancing for LD Notations Given S, T of source resp. target resources Divides S, T such that k i=1 Si = S and l j=1 Tj = T Determines (Si , Tj ) whose elements are to be compared The idea of load balancing is to distribute the computation of Si × Tj over m processors Task τ: Comparing elements in (Si , Tj ) c(τ) = |Si | · |Tj | block B: Set of all tasks assigned to a single processor c(B) = t∈B c(τ) MSE = m i=1 c(Bi ) − m j=1 c(Bj ) m 2 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
  • 16. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Load Balancing for LD Notations Given S, T of source resp. target resources Divides S, T such that k i=1 Si = S and l j=1 Tj = T Determines (Si , Tj ) whose elements are to be compared The idea of load balancing is to distribute the computation of Si × Tj over m processors Task τ: Comparing elements in (Si , Tj ) c(τ) = |Si | · |Tj | block B: Set of all tasks assigned to a single processor c(B) = t∈B c(τ) MSE = m i=1 c(Bi ) − m j=1 c(Bj ) m 2 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
  • 17. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Load Balancing for LD Notations Given S, T of source resp. target resources Divides S, T such that k i=1 Si = S and l j=1 Tj = T Determines (Si , Tj ) whose elements are to be compared The idea of load balancing is to distribute the computation of Si × Tj over m processors Task τ: Comparing elements in (Si , Tj ) c(τ) = |Si | · |Tj | block B: Set of all tasks assigned to a single processor c(B) = t∈B c(τ) MSE = m i=1 c(Bi ) − m j=1 c(Bj ) m 2 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 8/30
  • 18. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Running Example 4 tasks {τ1, τ2, τ3, τ4} Respective complexities {7, 1, 8, 3} 2 processors P1, P2 Tasks: 7 1 8 3 Processors: 1 1 τ1 assigned to P1: 7 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 9/30
  • 19. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Na¨ıve Load Balancer Idea Divides tasks between processors based on their index and regardless of complexity Example Processors assignment: 7 1 8 3 MSE = 30.25 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 10/30
  • 20. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Greedy Load Balancer Idea 1 Sorts tasks in descending order based on their complexity 2 Starting from the most complex task, assigns tasks to processors in order Example 1. Sorted tasks: 8 7 3 1 2. Processors assignment: 8 7 3 1 MSE = 2.25 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 11/30
  • 21. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Pair-Based Load Balancer Idea 1 Sorts tasks according to task complexity 2 In order, assigns ith and (n − i + 1)th tasks to Pi Example 1. Sort tasks: 1 3 7 8 2. Processors assignment: 1 3 7 8 MSE = 0.25 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 12/30
  • 22. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Particle Swarm Optimization Idea Initialization Best Known Positions (BKP) BKP ← Partitions the n tasks to m task blocks Computes fitness function F to the initial BKP F is the complexity difference between the most and least loaded blocks Initializes Best Known Fitness (BKF) to F Until a termination criterion is met Performs the particles migration, based on random particle velocity Recomputes F If F < BKF, updates BKF and BKP Return BKP Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 13/30
  • 23. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Particle Swarm Optimization Idea Initialization Best Known Positions (BKP) BKP ← Partitions the n tasks to m task blocks Computes fitness function F to the initial BKP F is the complexity difference between the most and least loaded blocks Initializes Best Known Fitness (BKF) to F Until a termination criterion is met Performs the particles migration, based on random particle velocity Recomputes F If F < BKF, updates BKF and BKP Return BKP Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 13/30
  • 24. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Particle Swarm Optimization Idea Initialization Best Known Positions (BKP) BKP ← Partitions the n tasks to m task blocks Computes fitness function F to the initial BKP F is the complexity difference between the most and least loaded blocks Initializes Best Known Fitness (BKF) to F Until a termination criterion is met Performs the particles migration, based on random particle velocity Recomputes F If F < BKF, updates BKF and BKP Return BKP Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 13/30
  • 25. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Particle Swarm Optimization Idea Initialization Best Known Positions (BKP) BKP ← Partitions the n tasks to m task blocks Computes fitness function F to the initial BKP F is the complexity difference between the most and least loaded blocks Initializes Best Known Fitness (BKF) to F Until a termination criterion is met Performs the particles migration, based on random particle velocity Recomputes F If F < BKF, updates BKF and BKP Return BKP Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 13/30
  • 26. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Particle Swarm Optimization Example Termination criterion: max number of iterations of 1 Initialization: 7 1 8 3 BKF = F = 11 First iteration: 7 1 8 3 F = 5 As F < BKF, updates BKF and PKB MSE = 6.25 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 14/30
  • 27. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Deterministic PSO (DPSO) DPSO PSO is non-deterministic PSO depends on a random selection of velocity for moving particles We propose the Deterministic PSO (DPSO) Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 15/30
  • 28. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Deterministic PSO (DPSO) Idea Partitions the n tasks to m task blocks Until termination criterion is met Finds the most overloaded block B+ and the least underloaded block B− Sorts tasks within B+ based in their complexities As far as a better balancing between B+ to B− is met Moves tasks in order from B+ to B− (task migration) Computes fitness function as complexity difference between B+ and B− Returns best known blocks Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 16/30
  • 29. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Deterministic PSO (DPSO) Why is DPSO deterministic? Only moves tasks from B+ to B− (no random migration) Sorts B+ tasks before task migration start Insures optimal load balancing between B+ and B− Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 17/30
  • 30. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Deterministic PSO (DPSO) Example Termination criterion: max number of iterations of 1 Initialization: B+ = 8 7 , B− = 1 3 F = 11 First iteration: Sorted B+ = 7 8 , B− = 1 3 Task migration: B+ = 8 , B− = 1 3 7 F = 3 MSE = 2.25 Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 18/30
  • 31. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Outline 1 Motivation 2 Load Balancing Approaches 3 Evaluation 4 Conclusion and Future Work Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 19/30
  • 32. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Evaluation Setup Orchid The parallel task generation was based on Orchid Orchid partitions the surface of the planet A task is the comparison of all points in two partitions Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 20/30
  • 33. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Evaluation Setup Datasets 1 5 synthetic geographic datasets Polygons’ sizes varied from 1 to 10 points 2 3 real geographic datasets Nuts DBpedia LinkedGeoData Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 21/30
  • 34. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Evaluation Setup Hardware 64-core server running OpenJDK 64-Bit Server 1.6.0 27 on Ubuntu 12.04.2 LTS. 8 quad-core processor Intel(R) Core(TM) i7-3770 CPU @ 3.40 GHz with 8192 KB cache Each experiment was assigned 20 GB of memory PSO PSO ran 5 times in each experiment and provide the mean of the 5 runs Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 22/30
  • 35. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Orchid vs. Parallel Orchid Goal Evaluate the speedup gained by using parallel Orchid For Nuts, PSO and DPSO up to 3 times faster For LinkedGeoData, PSO and DPSO up to 10 times faster 2 4 8 Number of threads 0.0 0.1 0.2 0.3 0.4 Time(min.) Naïve Greedy PairBased PSO DPSO Orchid Nuts runtime 2 4 8 Number of threads 1 10 100 1000 Time(min.) Naïve Greedy PairBased PSO DPSO Orchid LinkedGeoData runtime Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 23/30
  • 36. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Orchid vs. Parallel Orchid Goal PSO and DPSO are capable of achieving superlinear performance, as processor caches are faster than RAM Greedy and pair-based fail to achieve even the run time of the normal Orchid, due to significant sorting time 2 4 8 Number of threads 0.0 0.1 0.2 0.3 0.4 Time(min.) Naïve Greedy PairBased PSO DPSO Orchid Nuts runtime 2 4 8 Number of threads 1 10 100 1000 Time(min.) Naïve Greedy PairBased PSO DPSO Orchid LinkedGeoData runtime Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 24/30
  • 37. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Parallel Load balancing Algorithms Goals Measure each algorithm runtime Qualify each algorithm data distribution using MSE 2 4 8 Number of threads 0.0 0.1 0.2 0.3 0.4 Time(min.) Naïve Greedy PairBased PSO DPSO Orchid Nuts runtime 2 4 8 Number of threads 1010 1011 1012 1013 MSE Naïve Greedy PairBased PSO DPSO Nuts MSE Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 25/30
  • 38. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Parallel Load balancing Algorithms Goals Measure each algorithm runtime Qualify each algorithm data distribution using MSE 2 4 8 Number of threads 0.0 0.1 1.0 10.0 Time(min.) Naïve Greedy PairBased PSO DPSO Orchid DBpedia runtime 2 4 8 Number of threads 1011 1012 1013 MSE Naïve Greedy PairBased PSO DPSO DBpedia MSE Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 26/30
  • 39. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Parallel Load balancing Algorithms Goals Measure each algorithm runtime Qualify each algorithm data distribution using MSE 2 4 8 Number of threads 1 10 100 1000 Time(min.) Naïve Greedy PairBased PSO DPSO Orchid LinkedGeoData runtime 2 4 8 Number of threads 1009 1010 1011 MSE Naïve Greedy PairBased PSO DPSO LinkedGeoData MSE Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 27/30
  • 40. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Outline 1 Motivation 2 Load Balancing Approaches 3 Evaluation 4 Conclusion and Future Work Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 28/30
  • 41. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Conclusion and Future Work Conclusion Presented load balancing techniques for link discovery Proposed deterministic PSO (DPSO) Combined load balancing algorithms with Orchid Evaluated on real and artificial datasets Future Work Enable splitting of one task over multiple processors Implement a caching techniques Study the combination of DPSO with other task generation algorithms Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 29/30
  • 42. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Conclusion and Future Work Conclusion Presented load balancing techniques for link discovery Proposed deterministic PSO (DPSO) Combined load balancing algorithms with Orchid Evaluated on real and artificial datasets Future Work Enable splitting of one task over multiple processors Implement a caching techniques Study the combination of DPSO with other task generation algorithms Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 29/30
  • 43. Motivation Load Balancing Approaches Evaluation Conclusion and Future Work Thank You! Questions? Mohamed Sherif Augustusplatz 10 D-04109 Leipzig sherif@informatik.uni-leipzig.de http://aksw.org/MohamedSherif http://aksw.org/Projects/LIMES #akswgroup Mohamed Ahmed Sherif and Axel-Cyrille Ngonga Ngomo — DPSO 30/30