10
Network Resource Management Jason Gaedtke Chief Scientist W3C Video on the Web Workshop December 2007

Network Resource Management Jason Gaedtke Chief Scientist W3C Video on the Web Workshop December 2007

Embed Size (px)

Citation preview

Page 1: Network Resource Management Jason Gaedtke Chief Scientist W3C Video on the Web Workshop December 2007

Network Resource ManagementNetwork Resource ManagementJason GaedtkeChief Scientist

W3C Video on the Web WorkshopDecember 2007

Page 2: Network Resource Management Jason Gaedtke Chief Scientist W3C Video on the Web Workshop December 2007

04/18/23 © Cable Television Laboratories, Inc. 2007. All Rights Reserved. Proprietary/Confidential 2

TopicsTopics

• Abstract Compelling network neutrality arguments notwithstanding, not

all IP traffic exhibits uniform distribution requirements (e.g., bandwidth, latency, jitter and TTL).

Further, automated P2P file-sharing agents exploit TCP congestion control algorithms to gain a disproportionate share of network resources.

Some measures should be explored to address this natural, shared-network heterogeneity.

• Heterogeneous Applications and Network Requirements• Web Video Distribution Trends• A Resource Consumption Example (Briscoe Draft)• Potential Management Strategies• References and Collaborative Activities

Page 3: Network Resource Management Jason Gaedtke Chief Scientist W3C Video on the Web Workshop December 2007

04/18/23 © Cable Television Laboratories, Inc. 2007. All Rights Reserved. Proprietary/Confidential 3

Heterogeneous Apps and Network ReqsHeterogeneous Apps and Network Reqs

• Real-Time Apps: dependent upon low-latency delivery, exhibit highly-variable bandwidth requirements, few simultaneous connections Online gaming VoIP and video chat IM and Presence Streaming video

• Interactive Services: tolerant of modest delivery delays, modest bandwidth, few simultaneous connections E-mail Web browsing Progressive download

• Content Distribution: automated, many simultaneous connections, greedy – will consume available bandwidth P2P file-sharing File/mail/news/Web servers

Page 4: Network Resource Management Jason Gaedtke Chief Scientist W3C Video on the Web Workshop December 2007

04/18/23 © Cable Television Laboratories, Inc. 2007. All Rights Reserved. Proprietary/Confidential 4

Web Video Distribution (Background)Web Video Distribution (Background)

• Web Servers/Farms Simple client/server architecture Commodity servers, scaled horizontally

– Capacity– Redundancy/Availability

• Content Distribution Networks (CDNs) Specialized client/server architecture with aggressive caching Geographical distribution and load-balancing

• P2P Networks Decentralized, distributed and self-organizing “Super-nodes” avoid n2 link scaling and search Participants contribute bandwidth, storage and processing

• Hybrid CDN/P2P Networks Benefits of P2P resource sharing; <10% distro costs Seed and “long-tail” content sourced via CDN caches

Page 5: Network Resource Management Jason Gaedtke Chief Scientist W3C Video on the Web Workshop December 2007

04/18/23 © Cable Television Laboratories, Inc. 2007. All Rights Reserved. Proprietary/Confidential 5

A Resource Consumption Example (Briscoe, draft-briscoe-tsvwg-relax-fairness)A Resource Consumption Example (Briscoe, draft-briscoe-tsvwg-relax-fairness)

• 10Mbps, shared access network, 100 subscribers 80 subscribers primarily interactive Web/e-mail:

– 10% concurrency, 2 TCP connections each– 9.9kbps average during congestion– 7.1MB per day (16-hours active)

20 automated P2P file-sharing clients: – 100% concurrency, 100 TCP connections each– 496kbps average during congestion– 3.6GB per day – 500:1 volume skew

TCP congestion control treats each flow equally; greedy apps spawn many connections

Page 6: Network Resource Management Jason Gaedtke Chief Scientist W3C Video on the Web Workshop December 2007

04/18/23 © Cable Television Laboratories, Inc. 2007. All Rights Reserved. Proprietary/Confidential 6

Resource Consumption (4x Capacity)Resource Consumption (4x Capacity)

• 40Mbps shared, 100 subscribers 80 interactive Web/e-mail:

– 4% active (due to more responsive apps)– 40kbps (vs. ~10kbps) during congestion– 11MB (vs. ~7MB) per day

20 automated P2P file-sharing: – 2Mbps (vs. ~500kbps) during congestion– 14GB (vs. ~3.5GB) per day

As expected, a 4x increase in network capacity yields a 4x increase in average, per-flow rates under congestion; only exacerbates skew (>1250:1)

Page 7: Network Resource Management Jason Gaedtke Chief Scientist W3C Video on the Web Workshop December 2007

04/18/23 © Cable Television Laboratories, Inc. 2007. All Rights Reserved. Proprietary/Confidential 7

Resource Consumption (Capacity + Churn)Resource Consumption (Capacity + Churn)

• 40Mbps shared, 60 subscribers: 50 interactive Web/e-mail (30 churn):

– 2.5% active– 80kbps (vs. 40kbps) during congestion– 14MB (vs. 11MB) per day

10 automated P2P file-sharing (10 churn): – 4Mbps (vs. 2Mbps) during congestion– 29GB (vs. 14GB) per day

Trends: fewer subscribers, greater network capacity/cost, >2000:1 consumption skew; ergo,

rational operators will not add capacity

Page 8: Network Resource Management Jason Gaedtke Chief Scientist W3C Video on the Web Workshop December 2007

04/18/23 © Cable Television Laboratories, Inc. 2007. All Rights Reserved. Proprietary/Confidential 8

Resource Consumption SummaryResource Consumption Summary

• TCP congestion control treats all flows equally• Automated P2P agents are (very) greedy

100+ simultaneous connections 100% concurrency

• These aggressive algorithms will absorb an increasing amount of added capacity, thus degrading cost/benefit for other users

• Light, interactive users subsidize P2P distribution

• New economic/technical management strategies should be explored

Page 9: Network Resource Management Jason Gaedtke Chief Scientist W3C Video on the Web Workshop December 2007

04/18/23 © Cable Television Laboratories, Inc. 2007. All Rights Reserved. Proprietary/Confidential 9

Potential Management StrategiesPotential Management Strategies

• Upstream/downstream rate limiting• Aggregate capacity limiting (tiering)• Application-specific throttling (via DPI)• Differentiated/priority service classes• Reservation-based resource management• Explicit protocol-level feedback/heuristics• Variable/metered pricing strategies• Bandwidth/resource trading schemes and

virtual economies• Others?

Page 10: Network Resource Management Jason Gaedtke Chief Scientist W3C Video on the Web Workshop December 2007

04/18/23 © Cable Television Laboratories, Inc. 2007. All Rights Reserved. Proprietary/Confidential 10

References and Collaborative ActivitiesReferences and Collaborative Activities

• IETF Transport Area RFC 2309: Recommendations on Queue Management and

Congestion Avoidance RFC 2581: TCP Congestion Control RFC 2914: Congestion Control Principles draft-briscoe-tsvwg-relax-fairness

• DCIA P4P Working Group• CableLabs PacketCable Multimedia QoS• DSL Forum TR 58/59• Harvard SEAS and Tribler.org

Bandwidth Virtual Economy