Dynamic Resource Allocation in Clouds: Smart Placement with Live

MigrationMakhlouf Hadji

Ingénieur de Recherchemakhlouf.hadji@irt-systemx.fr

FONDATION DE COOPERATION SCIENTIFIQUE

2

NOS VALEURS

3

❶ Excellence

- Talents- Résultats- Echanges

❷ Souplesse- Fonctionnement- Partenariat- Projets ❸ Rigueur

- Propriété Intellectuelle- Confidentialité- Exécution

PROGRAMMES DE R&D

4

Ingénierie numérique des Systèmes et Composants

PROJETS R&D Projets opérationnels depuis le lancement

5

Projets Opérationnels

M€ Financement Industriel

915,5

ETP/an sur 3 ans

Partenaires Industriels

Partenaires Académiques

1104212

Thèses29

I- Smart Placement in Clouds

Smart Placement in Clouds

7

Define the best strategy to place VMs workloads leading to optimally reduce infrastructure costs.

ESX 1 ESX 2… ESX N

Infrastructure servers

???

Cloud End-UsersVMs demand management

VMs placement strategy

Problem: Based on allocating and hosting N VMs on a physical infrastructure of X Serveurs, what is the best manner to optimally place workloads to minimize different infrastructure costs ?

Benefits Resources optimization, Minimization of infrastructure costs, Energy consumption optimization.

VM Placement problem

ESX 1 ESX 2… ESX N

Non optimalplacement

ESX 1 ESX 2… ESX N

Optimal placement

Challenges of the problem: Exponential number of cases to enumerate.

8

Smart Placement in Clouds

French Providers Point of View

9

Smart Placement in Clouds

ESX 1 ESX 2… ESX N

Infrastructure serveurs

Utilisateurs des services Cloud

Gestion de la demande de VM

Forcasting & Scheduling

small

small

mediumlarge

Due to fluctuations in users’ demands, we use Auto-Regressive (AR(k)) process, to handle with future demands:

Problem Complexity :NP-Hard Problem: One can construct easily a plynomial reduction from the NP-Hard notary problem of the Bin-Packing.

tit

k

iit dd

1

10

Smart Placement in CloudsMathematical formulation:

Formulation as ILP:The corresponding mathematical model is an Integer Linear Programming: difficulties to characterize the convex hull of the considered problem and the optimal solution.

else. 0

iserver in hosted is VM if 1

,,

,

,1,,

:ToSubject

min

1

1 1 1 1

jij

ij

j

N

iij

ijijij

N

i

I

j

N

i

I

jijjijij

y

jiNx

Ijdx

NiIjyCx

xPyZ

11

Smart Placement in CloudsMinimum Cost Maximum Flow Algorithm

S T

(1; 0,33)

(2; 0,23)

(1; 0,14)

(1; 0)

(2; 0)

(1; 0)

(Di; 0

)(Di; 0)

Instance i

Legend: (capacity; cost)

12

Smart Placement in CloudsMinimum Cost Maximum Flow Algorithm

S T

(1; 0,33)

(2; 0,23)

(1; 0,14)

(1; 0)

(2; 0)

(1; 0)

(D1; 0

)(D1; 0)

Small Instance

(0; ∞)

(2; 0,23)

(1; 0,14)

(0; 0)

(2; 0)

(1; 0)

Medium Instance

(D2; 0) (D2; 0)

13

Smart Placement in CloudsSimulation Tests: Case of (0;1) Random Costs

We consider (0; 1) Random hosting costs between each couple of vertices (a, b), where a is a fictif node, and b is a physical machine (server).

Random Hosting Costs Scenario

14

Smart Placement in CloudsSimulations Tests: Case of Inverse Hosting Costs:

Inverse Hosting Costs Scenario

We consider inversed hosting costs function between each couple of vertices (a, b), where a is a fictif node, and b is a physical machine:

Where represents the available capacity on the considered arc. est une fonction non nulle.

ababab

ab gCCf

g otherwise ,0 if )(

1

abC f

II- Live Migration of VMs

Live Migration of VMs

Migration process:

16

ESXXen

KVM Hyper-V O

FF

Polytops, faces and facets

facetsface

jjxcMin

m1ibxa ijij ,...,,Subject To constraints:

nixi ,...,1,1,0

*x

Live Migration of VMs

Polyhedral Approachs

11 x

22 x

20

18

Live Migration of VMs

Decision Variables: if A VM k is migrated from i to j (0 else). Prevent backword migration of a VM:

Server’s destination uniqueness of a VM migration:

Servers’ power consumption limitation constraints:

Etc…

Polyhedral Approachs

Some Valid Inequalities of our Problem:

1ijkZ

1' jlkijk ZZ

1'

,1

m

ijjijkZ

jcurrentjjijk

m

ik

qi

k

yppzp

1,max,

'

1 1

19

Live Migration of VMsPolyhedral Approachs

otherwise 0idle is server if 1

otherwise 0j, toi from migrated is VM if 1

'

1

1

',,',1,,1',,1,',,,1,1

:ToSubject

'max

0

'

1 max,

'

1,

'

1 1

'

1 1,max,

'

,1

'

'

1

'

1

'

1 1,

iy

z

Ttz

P

Pmy

yqz

yPPzp

z

kkjlmlqkqkmijmizz

zpyPM

i

kijk

kijk

m

i j

m

jcurrentj

i

m

jii

q

kijk

m

i

q

kjcurrentjjijkk

m

ijjijk

jijlkijk

m

i

m

i

m

j

q

kijkkiidli

i

i

i

20

Live Migration of VMsPolyhedral Approachs

Number of used servers when taking into account Migrations

21

Convergence Time (in seconds) of Migration Algorithm:

22

Live Migration of VMsPolyhedral Approachs

Percentage of Gained Energy when Migration is Used

5 10 15 20 25 30

10 35,55 36,59 00 00 00 00

20 27,29 34,00 35,23 38,50 00 00

30 17,48 27,39 35,21 40,32 41,89 36,65

40 16,77 18,85 22,02 32,31 39,90 40,50

50 10,86 16,17 19,85 22,30 39,20 36,52

60 08,63 14,29 18,01 22,13 25,15 30,68

70 08,10 14,00 14,86 15,90 22,91 23,20

80 07,01 10,20 10,91 15,34 17,02 21,60

90 06,80 09,32 10,31 14,70 16,97 19,20

100 05,90 07,50 08,40 12,90 16,00 14,97

Thank you