Copy of first presentation (by aj. boom)

presented by

Nay Myo Sandar (Chan Chan)

Shinawatra University

Data Placement in Cloud Computing using Linear Programming

Outline1. Overview of Data Placement in Cloud Computing

2. Cloud Computing for Data Placement

• Introduction to Cloud Computing

• Benefits of Cloud Computing for Data Placement

3. Data Placement Problems in Cloud

4. Proposed Solution for Data Placement in Cloud Computing

• Why Linear Programming is a good solution

5. Applications of using Linear Programming

6. Example:Farming Problems

7. Study Plan (For this semester)

8. Summary

1. Overview of Data Placement in Cloud Computing

• Data placement = data locality management in cloud computing

• Local (in-house) resources are not sufficient

• Cloud computing is an efficient solution for data placement

• In this study, we will propose a solution to select data centers/cloud providers to store datao faster data accesso faster computation o longer data availabilityo cheaper computation & storage

2. Cloud Computing for Data Placement

Introduction to Cloud Computing• Cloud computing supply computing

resources/services (e.g., networks, servers, storage, applications and services) through the Internet

• 3 types of cloud Public cloud - sells services to anyone (e.g.,

Amazon, GoGrid, and Microsoft) Private cloud - services are limited to a limited

number of people Hybrid cloud - composition of public cloud and

private cloud

Benefits of Cloud Computing

Reliability (Service Level Agreement / SLA)

• agreement between the provider and the customer

• make guarantee data availability

Accessibility

• store files safely in the infrastructure

• can access personal files or information at any computer, anytime and anywhere

Benefits of Cloud Computing(Continued)

Variety

• offer many services

Scalability

• give ability to service nearly unlimited storage capacity for a great number of data

• can easily scale up or down the amount of resources by limiting them according to user requirements

Benefits of Cloud Computing(Continued)

Cost-effectiveness

• pay per real usage i.e., pay-per-use option

• pay resources with fixed (and cheaper) price (i.e., reservation option)

• bid resources (i.e., spot option)

3. Data Placement Problems in Cloud

Size of data

• data becomes terabytes or exabytes

• limited local storage

Monetary cost for data placement/computation

• pay several other costs - network bandwidth, data storage, maintenance, data durability guarantee, servers, services, software, licenses etc.

3. Data Placement Problems in Cloud(Continued)

Resource reliability

• cannot fully guarantee that resources are fully reliable to store our data

Network bandwidth

• transfer big data through unreliable network can be unacceptable delay moving certain data

Geographical data movement

• placing data in different continents might not be trivial


Data sensitivity

• sensitive information, e.g., customers' credit card numbers


Data availability

• cannot get data anywhere and anytime

• if the service provider's server crashed, so does our files

Data integrity

• multiple replicas of the same data should be consistent


• Apply linear programming to obtain the optimal solution for data placement in cloud computing

• Define an objective / multiple objectiveso maximize processing speedo minimize data access delayo minimize monetary costo maximize safety / minimize security attacks

• Define constraints o to control users' budgetso to control the available storage spaceo to meet SLAso to guarantee minimum data access delayo to govern the number of data backups

etc.


Why Linear Programming is a good solution

• linear programming is mathematical programming

• can achieve the optimal outcome

• can be practically solved by personal computers

5. Applications using Linear Programming

• Transportation

• Agriculture

• Warfare

• Telecommunications

• Microchips

etc.

6. Example: Farming ProblemsA farmer

• has 10 acres to plant in wheat and rye

• has to plant at least 7 acres

• has only $1200 to spend

• has each acre of wheat costs $200 to plant and each acre of rye costs $100 to plant

• has to get planting done in 12 hours

• takes an hour to plant an acre of wheat and 2 hours to plant an acre of rye

If the profit is $500 per acre of wheat and $300 per acre of rye, how many acres of each should be planted to maximize profits?

6. Example: Farming Problems(Continued)

Steps

1. Read the whole problem

2. Define variables

3. Express the objective function and the constraints

4. Graph the constraints

5. Find the corner points to the region of feasible solutions

6. Evaluate the objective function at all the feasible corner points

6. Example: Farming Problems(2) Define variables

let x = the number of acres of wheat

and y = the number of acres of rye

6. Example: Farming Problems(3) Express the objective function and

the constraints

First, the objective = to maximize profits

current profit of each acre of wheat = $500

current profit of each acre of rye = $300

Profit = 500 x+300 y

6. Example:Farming Problems(Continued)

• acreage constraint

• cost constraint

• time constraint

x + y <= 10

x + y >= 7

200 x + 100 y <=1200

x + 2y <= 12

implied constraints

x>=0 and y>=0

acreage

cost

time

6. Example: Farming Problems(4) Graph the Constraints

First we find the intercepts

Intercepts x y

acreage x + y<=10 10 10

x + y>=7 7 7

cost 200 x + 100 y<=1200 6 12

time x + 2y<=12 12 6

6. Example: Farming ProblemsGraph of acreage constraint

x + y <= 10 x + y <= 10

x + y >= 7

6. Example: Farming ExampleGraph of cost constraint

x intercept of 6 and y intercept of 12

200x + 100y <= 1200

6. Example: Farming ProblemsGraph of time constraint

x intercept of 12 and y intercept of 6

x + 2y <= 12

6. Example: Farming Problems(5) Find the feasible corner points

cost constraint equation2nd acreage constraint

cost constraint equationtime constraint equation

time constraint equation2nd acreage constraint


(Continued)

200 x + 100 y = 1200 200x + 100y = 1200

A

x + y = 7 (-100) -100x-100y = -700

x = 5

substitute into x + y = 7 and get y = 2

so A = (5,2)


(Continued)

200x + 100y = 1200 200x + 100y =1200

B

x + 2y = 12 (-50) -50x - 100y = -600

x = 4

substitute into x + 2y = 12 and get y = 4

so B = (4,4)


(Continued)

x + 2y = 12

C

x + y = 7

subtract the two equations and get y = 5 and x = 2

so C = (2,5)

6. Example: Farming Problems(6) Evaluate the objective function in all

of the feasible corner points

A=(5,2) Profit = $500(5)+$300(2) = $3100

B=(4,4) Profit = $500(4)+$300(4) = $3200

C=(2,5) Profit = $500(2)+$300(5) = $2500

The winner is B.

The maximize the profit of $3200 is obtained by planting 4 acres of each crop.

Study Plan (For this semester)

7. Summary

• place data into specific locations (both private/public clouds) effectively and efficiently

• cloud computing can give many benefits to users

• data placement in cloud problems need to be handled

• linear programming can give the feasible solutions for the data placement problems

THANK YOU

Copy of first presentation (by aj. boom)

Technology