1.on-line algorithms 2.energy efficient utilization of resources in cloud

36
1. On-Line Algorithms 2. Energy efficient utilization of resources in cloud Raziel Hess-Green 1

Upload: linore

Post on 25-Feb-2016

31 views

Category:

Documents


3 download

DESCRIPTION

1.On-Line Algorithms 2.Energy efficient utilization of resources in cloud. Raziel Hess-Green. On-Line Algorithms. A small intro Raziel Hess-Green. Elevator or S tairs problem. More known as: “ski-rental problem” Stairs: takes time S Elevator: takes time L

TRANSCRIPT

Page 1: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

1

1. On-Line Algorithms

2. Energy efficient utilization of resources in cloud

Raziel Hess-Green

Page 2: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

2

On-Line AlgorithmsA small intro

Raziel Hess-Green

Page 3: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

3

More known as: “ski-rental problem”

Stairs: takes time S Elevator: takes time L<S

The ultimate question: How long to wait?

Elevator or Stairs problem

Page 4: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

4

Competitive ratio – Alg/OPT◦ worst case over all possible events◦ Alg = cost of algorithm◦ OPT = optimal cost in hindsight.

Evaluate on-line algorithms

Page 5: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

5

Wait until elevator comes◦ What if it’s broken?

Take stairs immediately◦ Bad competitive ratio - S/L

Back to elevators and stairs

Page 6: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

6

Wait until you should have taken the stairs, then take the stairs

Case 1: ◦ Elevator comes before time S-L: optimal.

Case 2: ◦ Elevator comes after: you paid 2S-L, OPT paid S.

Ratio = 2 - L/S.

2-competitive

Page 7: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

7

Elevator arrives right after you give up: ◦ If you wait longer,

numerator goes up but the denominator stays the same, so your ratio is worse.

◦ If you wait less, then the numerator and the denominator go down by the same amount, worse.

That’s the best possible

Page 8: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

8

BP:◦ Given N items with sizes s1, s2,…, sN, where 0 si

1. The bin packing is to pack these items in the fewest bins, given that each bin has unit capacity.

On-line bin packing:◦ Each item must be placed in a bin before the size

of the next item is given.

Stay tuned for more..

Bin Packing

Page 9: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

9

Page 10: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

10

Energy efficient utilization of resources in cloud

computing systemsYoung Choon Lee, Albert Y. Zomaya

Page 11: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

11

2000 – 2005◦ Doubled!◦ 2005 cost 7.2 bn US$

2005-2010◦ Predicted by the EPA at 2007 to double again◦ Actually added around 56% (J. Koomey)

Mainly due to 2008 recession 2011

◦ 2% of USA electricity

Elictricity in Data Centers

Page 12: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

12

Electricity Bill With Great Power Comes Huge:

Page 13: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

13

Cloud Computing allows for fuller utilization of hardware

Energy consumption is turning into a major issue Costly CO2 emission

Must hold enough resources to handle peak demand

Energy grows linearly with utilization

Utility Computing

Page 14: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

14

20% utilization Idle servers can use 60% of full utilization Turning off is problematic

◦ Long turn on time◦ May increase failure rate

Turn Off Power?

Page 15: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

15

Must have the server totally unutilized to enable sleep mode

Dynamic Voltage and Frequency Scaling (DVFS)◦ Intel SpeedStep◦AMD PowerNow!

Started in laptops and mobile devices Now used in servers Much more research on this:

◦ PowerNap (ASPLOS ’09)

Power Saving Mode

Page 16: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

16

Cloud Application Energy

Model

Page 17: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

17

Resources◦ set R of r resources/processors

fully interconnected Homogeneous

◦ Communication◦ Same DC

Live Migration

Cloud Model

Page 18: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

18

IaaS, SaaS or PaaS regarded as tasks Assumed: known time and CPU demand

◦ IaaS has predefined time/CPU requirements◦ For SaaS and PaaS- obtain estimates from history

and/or from consumer

Application Model

Page 19: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

19

linear relationshipwith processing time and utilization:

◦ - utilization of task on

Energy during Power Save mode:

Energy Model

jt

Page 20: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

21

Assigning a set N of n tasksto a set R of r cloud resources

Maximize resource utilization◦ In order to minimize energy consumption◦ By enabling resources to sleep

Without violating constraints◦ time◦ Usage◦ Hard constraints

Task consolidation problem

Page 21: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

22

Two algorithms presented, differ only in cost function

ECTC ◦ Explicitly computes energy consumption

MaxUtil◦ Average utilization -

during processing time of the task to schedule◦ Increase consolidation density

The Algorithms

Page 22: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

23

The Algorithm:

Page 23: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

24

τ0 – ((τ1 +.τ2)

◦ - utilization rate of the task ◦ - total processing time of the task◦ τ1- time task will run alone◦ τ2- time task will run in parallel

ECTC

Page 24: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

25

Maximize average consolidation density◦ Over all processing time of task j

MaxUtil0

1,

0

i

i j

Uf

Page 25: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

26

Example ECTC

Page 26: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

27

Example MaxUtil

Page 27: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

28

Random ECTC MaxUtil

1,500 experiments◦ 50 different number of tasks

100-5,000 with intervals of 100◦ 10 mean inter-arrival times (10 -100)

Poisson process

Experimental evaluation

Page 28: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

29

Three usage patterns◦ Random

Uniformly distributed between 0.1 and 1◦ Low

Gaussian, mean utilization rates of 0.3◦ High

Gaussian, mean utilization rates of 0.7

Usage patterns

Page 29: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

30

Task processing time◦ Exponential distribution

◦ Assume: 300-200 watt active mode consumption

_m◦ Adding migration

Page 30: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

31

Relative energy savings◦ MaxUtil◦ ECTC

Different resource usage patterns◦ Low◦ High◦ Random

Results

Page 31: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

32

MaxUtil and ECTC vs Random

Page 32: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

33

Low resource usage

Page 33: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

34

High resource usage

Page 34: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

35

Random resource usage

Page 35: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

36

Important problem Strict modeling

◦ All demands known exactly (time, usage)◦ Communication is “free”

And yet: No sophisticated algorithms No “make sense” for results No comparing to previous work

◦ “existing task consolidation algorithms are not directly comparable to our heuristics”

Ending Remarks

Page 36: 1.On-Line Algorithms 2.Energy  efficient  utilization  of  resources  in cloud

37

Stochastic Bin Packing (SBP) problem◦ each virtual machine's bandwidth demand is

treated as a random variable. ◦ both offline and online versions are treated◦ assumption: VMs' bandwidth consumption obeys

normal distribution◦ show a 2-approximation algorithm for the offline

version ◦ (2+Ɛ)-competitive algorithm for online version

SBP David Breitgand, Amir Epstein (IBM, Haifa)