Optimizing Cost and Performance for Content Multihoming

OPTIMIZING COST AND PERFORMANCE FOR CONTENT MULTIHOMING

SIGCOMM’12-PIGGY, 2013.03.18

OUTLINE• What is Content Multihoming• Goal• Control Framework• Global Optimization• Local Adaptation• Evalution

CONTENT MULTIHOMING

CDN DIVERSITY

CDN DIVERSITY

CDN DIVERSITY

GOAL• Algorithms and protocols that optimize

• Content publisher cost• Content viewer performance

• A content object can be delivered from multiple CDNs, which CDN(s) should a content viewer use?

NOTATION

CONTROL FRAMEWORK

PASSIVE VS. ACTIVE CLIENT• Passive client

• Use one CDN edge server at a time• Active client

• Adaptation algorithm• Multiple CDN servers for a single content object

PROBLEM STATEMENT (Q)• QoE guarantee

• CDN k is providing the required features to deliver content object i

• exceeds the performance target• Cost optimization

• Balance load to multiple CDNs to minimize total cost

ACTIVE CLIENT• Virtual CDN

• Primary CDN• Backup CDN• k’ = (k, j)

COMPUTING OPTIMIZATION(CMO)• Problem Q has an optimal solution which

assigns a location object into a single CDN

• K|A|N

BASIC IDEA

EXTENSION • CDN subscription levels

• Fix fee to different usage levels• Different levels as an individual CDN

• Per-request cost• Extend vector dimension to R+1

• Multiple streaming rates• Independent content objects

LOCAL ADAPTATION• QoE protection• Prioritized guidance• Low session overhead

LOCAL ADAPTATION• Similar to TCP AIMD• Total workload control• Priority assignment

EVALUATION SETTING

COST SAVING

COST SAVING

ACTIVE CLIENT SETTING• Clients

• 500+ Planetlab nodes with Firefox 8.0 + Adobe Flash 10.1

• Two CDNs• Amazon CloudFront• CDN3

ACTIVE CLIENT TEST CASE

STRESS TESTS (STEP-DOWN)

STRESS TESTS (RAMP-DOWN)

STRESS TESTS (OSCILLATION)

ACTIVE CLIENT QOE GAIN

CONCLUSION• We develop and implement a two-level

approach to optimize cost and performance for content multihoming: • CMO: an efficient algorithm to minimize publisher cost

and satisfy statistical performance constraints• Active client: an online QoE protection algorithm to

follow CMO guidance and locally handle network congestions or server overloading

Q&A