360 likes | 372 Views
Discusses online algorithms for energy-efficient cloud resource utilization. Examines elevator or stairs problem and bin packing in online scenarios. Explores energy-efficient strategies and algorithms for cloud computing systems.
E N D
1. On-Line Algorithms2. Energy efficient utilization of resources in cloud Raziel Hess-Green
On-Line Algorithms A small intro Raziel Hess-Green
Elevator or Stairs problem • More known as: “ski-rental problem” • Stairs: takes time S • Elevator: takes time L<S • The ultimate question: • How long to wait?
Evaluate on-line algorithms • Competitive ratio – Alg/OPT • worst case over all possible events • Alg = cost of algorithm • OPT = optimal cost in hindsight.
Back to elevators and stairs • Wait until elevator comes • What if it’s broken? • Take stairs immediately • Bad competitive ratio - S/L
2-competitive • 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.
That’s the best possible • 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.
Bin Packing • BP: • Given N items with sizes s1, s2,…, sN, where0si 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..
Energy efficient utilization of resources in cloudcomputing systems Young ChoonLee, Albert Y. Zomaya
Elictricity in Data Centers • 2000 – 2005 • Doubled! • 2005 cost 7.2 bnUS$ • 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
With Great Power Comes Huge: Electricity Bill
UtilityComputing • 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
Turn Off Power? • 20% utilization • Idle servers can use 60% of full utilization • Turning off is problematic • Long turn on time • May increase failure rate
Power Saving Mode • 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)
Model • Cloud • Application • Energy
Cloud Model • Resources • set R of r resources/processors fully interconnected • Homogeneous • Communication • Same DC • Live Migration
Application Model • 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
Energy Model • linear relationshipwith processing time and utilization: • - utilization of task on • Energy during Power Save mode:
Task consolidation problem • 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
The Algorithms • 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
ECTC • 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
MaxUtil • Maximize average consolidation density • Over all processing time of task j
Experimental evaluation • 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
Usage patterns • 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
Task processing time • Exponential distribution • Assume: 300-200 watt active mode consumption • _m • Adding migration
Results • Relative energy savings • MaxUtil • ECTC • Different resource usage patterns • Low • High • Random
Ending Remarks • 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”
SBP David Breitgand, Amir Epstein (IBM, Haifa) • 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