51
Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1 , Zhiyi Huang 1 , Chuan Wu 1 , Zongpeng Li 2 , Francis C.M. Lau 1 1 The University of Hong Kong 2 University of Calgary

Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Embed Size (px)

Citation preview

Page 1: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Online Auctions in IaaS Clouds: Welfare and

Profit Maximization with Server Costs

Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1,

Zongpeng Li 2, Francis C.M. Lau 1

1 The University of Hong Kong2 University of Calgary

Page 2: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Outline

Background Problem settings Difficulties Mechanism design Performance evaluations Summary

Page 3: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Amazon EC2

Computing resources are packed into VMs by virtualization technology

Page 4: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Users demand for Cloud resources

Page 5: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Why auction

Users can determine their own VM types

Providers can price according to the current demand and supply relationship

Computing resources are limited

Page 6: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Amazon EC2 Spot Instances

Demand: a VM instance

Page 7: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Amazon EC2 Spot Instances

Demand: a VM instance

Page 8: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Our work

Our workTime-variant resource demands—— a customized VM

Start time

End time

Page 9: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

• S servers• R types of resources, with known capacities• T time slots• I users• For each user i: • Demands of user i:

Problem settings

Page 10: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

• Power consumption of each server can be formulated

Problem settings

Horizontal coordinate: resource(e.g., CPU) utilizationVertical coordinate: power consumption

Page 11: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

When Dynamic Voltage Frequency Scaling (DVFS)

server mode is enabled, .

Problem settings

Page 12: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Problem settings

Dynamic Voltage Frequency Scaling (DVFS) server mode is disabled.

Page 13: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Problem settings

Server cost is not considered.

Page 14: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

bid serv

erbid i on s ✔ =1 ✗ =0

time slot

resource

VM allocation model

Page 15: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

resource allocation constraint

server cost

VM allocation model

Page 16: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

1. y = allocated resource amount

2. No resource is over-provisioned

VM allocation model

Page 17: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

VM auction: NP-hardSever cost considerationResource reusability

Payment: For welfare maximization: it should be a threshold to filer out low-value bids For profit maximization: it should be as close to bidding price as possible

Challenges with online auction

Page 18: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Upon the arrival of bid i: 1. Use the current price to

calculate the payment on each server: payment(s)= 2. Choose the smallest payment and

the corresponding server . 3. If : Accept bid i. Serve bid i on

. Update . Update . Otherwise: Reject bid i .

Online auction for welfare maximization

Page 19: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Our online auction achieves:

How to update ?

Online auction for welfare maximization

Page 20: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Our online auction achieves: Truthfulness Polynomial time complexity High social welfare ---- competitive

ratio

How to update ?

Online auction for welfare maximization

Page 21: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Our online auction achieves: Truthfulness Polynomial time complexity High social welfare ---- competitive

ratio

How to update ?

Online auction for welfare maximization

Page 22: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Our online auction achieves: Truthfulness Polynomial time complexity High social welfare ---- competitive

ratio

How to update ? Using an online primal-dual framework

Online auction for welfare maximization

Page 23: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

An online primal-dual framework

Primal:

Dual:

Lagrange relaxation

Page 24: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

An online primal-dual framework

Primal:

Dual:

Lagrange relaxation

Resource

priceUtility of i

Page 25: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

An online primal-dual framework

Primal (P): Maximization problem Dual (D): Minimization problem P(or D): Primal(or Dual) objective

value under a feasible primal(or dual) solution

Weak duality: P<=P* <= D*<=D

Page 26: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

An online primal-dual framework

Primal (P): Maximization problem Dual (D): Minimization problem P(or D): Primal(or Dual) objective

value under a feasible primal(or dual) solution

Weak duality: P<=P* <= D*<=D

Competitive analysis: P>= (1/α)D

Tight the gap

Page 27: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

An online primal-dual framework

Competitive analysis: P>= (1/α)D Pi: The primal objective value after

dealing with bid i Di: The dual objective value after

dealing with bid i In order to get P>= (1/α)D, we

resort to satisfying Pi - Pi-1 >= (1/α) (Di – Di-1), given P0=D0=0.

Page 28: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

An online primal-dual framework

In order to guarantee Pi - Pi-1 >= (1/α) (Di – Di-1), we

have:

Solve for each i, r, s, t to minimize α.

Page 29: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

What are the goals of our online auction ?

Truthfulness Polynomial time complexity High social welfare ---- competitive ratio

How to update ? Using an online primal-dual framework, we solve , a function of .

Online auction for welfare maximization

Page 30: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

How to intuitively understand

Page 31: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Evaluation setup

Google Cluster Data contains information including resource demands (CPU, RAM, Disk), job arrival times and durations.

We translate each job into a VM bid, requesting R = 3 types of resources at the demands extracted from the traces (demand here is not much smaller than Crs ).

Default Parameters: Each time slot is 10 seconds, and a bid arrives every [1, 10]

time slot(s). The duration of each VM is between 10 and 3600 time slots.

Ur =50, Lr =1. hrs~[0.4, 0.6] for CPU, hrs~[0.005, 0.02] for RAM and disk. βrs~[1.7, 2.2] for CPU, βrs~[0.5, 1] for RAM and disk.

The capacity of each type of resource and the number of servers are roughly according to the total amount of demand from all bids multiplying a random number in [0.4, 0.8].

Page 32: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Performance evaluation of welfare maximization

Bar colors: difference with average # of time slots between the bid arrival time to its specified VM start time over all the bids.

Page 33: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Performance evaluation of welfare maximization

Page 34: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Performance evaluation of profit maximization

Bar colors: difference with the ratio between the upper and lower bound of bidding price per unit of resource demand

Page 35: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Summary

We design an online auction to do Cloud resource provisioning for users with time-variant demands.

We consider heterogeneous resources dynamically allocated, released, and reused.

We maximize social welfare with server cost which achieves a good competitive ratio.

We also apply an approach to maximize provider profit which achieves a good competitive ratio.

Page 36: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Thank you

Q&A

Page 37: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Backup Slides

Page 38: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Upon the arrival of bid i: 1. Using the current price of each type of resource

at each time slot within the resource execution duration to calculate the payment on each server:

payment(s)= 2. Choose the smallest payment among all s,

denoted by . The corresponding s is . 3. If : Accept bid i. Allocate resources on

to server bid i. Update . Update . Otherwise: Reject bid i.

Online auction for welfare maximization

Page 39: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

uiPrs(t)

relaxation Lagrange

Relaxation

Online primal-dual framework

Page 40: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

uiPrs(t)

relaxation

Online primal-dual framework

Page 41: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

uiPrs(t)

relaxation

utilitymarginal payment

Online primal-dual framework

Page 42: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

uiPrs(t)

relaxation

utilitymarginal payment

payment of bid i

Online primal-dual framework

Page 43: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

utilitymarginal payment

payment of bid i

Online primal-dual framework

Page 44: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

utilitymarginal payment

payment of bid i

1. Dual Feasibility P >= 1/α D

P*<=D*<=D P >= 1/α P*2. Complementary Slackness of KKT Optimality Condition:For any user i,

Main idea:

Online primal-dual framework

Page 45: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Intuitively, prs(yrs(t)) is a function of marginal cost under some predicted allocated amount.

Why ?Upon Bi’s arrival, for any t, r, s, prs(t)dir(t) > frs(yrs(t)+dir(t)) – frs(yrs(t))

≈ f ’rs(yrs(t))dir(t)

if

How to design prs(t)

Page 46: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Intuitively, prs(yrs(t)) is a function of marginal cost under some predicted allocated amount.

Why ?

Exponential function is growing fast!

if

How to design prs(t)

Page 47: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Pi – Pi-1 >= 1/α ( Di – Di-1 ), given P0 = D0 =0 PI >= 1/α DI α is the competitive ratio

How to design prs(t)

Page 48: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Pi – Pi-1 >= 1/α ( Di – Di-1 ), given P0 = D0 =0 PI >= 1/α DI α is the competitive ratio

How to design prs(t)

Page 49: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Pi – Pi-1 >= 1/α ( Di – Di-1 ), given P0 = D0 =0 PI >= 1/α DI α is the competitive ratio

How to design prs(t)

Page 50: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

Show the idea by drawing a picture on the whiteboard

How to design prs(t)

Page 51: Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Xiaoxi Zhang 1, Zhiyi Huang 1, Chuan Wu 1, Zongpeng Li 2, Francis C.M

>= bmax

Profit maximization