broadband internet performance: a view from the gateway srikanth sundaresan, walter de donato, nick...
TRANSCRIPT
Broadband Internet Performance: A View from the GatewaySrikanth Sundaresan, Walter de Donato,
Nick Feamster, Renata Teixeira, Sam Crawford, Antonio Pescapè
Georgia TechUniversity of Napoli
CNRS/UPMC Sorbonne UniversitésSamKnows
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
What Affects Broadband Performance?
• Important for regulators, consumers, ISPs• Notion of performance is fuzzy– What metrics should we measure?– How to measure them?
2
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Accurate Measurements are Difficult
End host measurements are not continuous, and affected by confounding factors
Home Network: AT&T DSL 6 Mbps Down, 512 Kbps Up
Last MileISP Network
speedtest.net: 4.4 Mbps, 140 Kbps
Netalyzr: 4.8 Mbps, 430 Kbps
3
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
The Case For the Gateway
Gateway enables periodic measurements, and can account for confounding factors
Home Network: AT&T DSL 6 Mbps Down, 512 Kbps Up
Last MileISP Network
speedtest.net: 4.4 Mbps, 140 Kbps
Netalyzr: 4.8 Mbps, 430 KbpsGateway: 5.6 Mbps, 460 Kbps
4
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Contributions
• Study of access networks from the gateway• Insight into how to benchmark performance– Effect of measurement technique, traffic shaping– Effect of access link properties, home equipment
• Insight into broadband performance– Consistency of performance– Loss, jitter, time-of-day effects
5
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
The Deployments
• Breadth: The FCC/SamKnows study– 4,000 gateways, 16 ISPs, multiple service plans
• Depth: The BISmark study– 16 gateways in Atlanta, on-demand measurements
• Duration: Dec 2010 – Jan 2011
SamKnows/BISmark
6
Last MileInternet
Nearby Server
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Results: Overview
• Throughput measurement technique depends on usage scenario
• Traffic shaping is highly variable across users• Access link characteristics affect performance• Modem buffers induce high latency
7
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Results: Overview
• Throughput measurement technique depends on usage scenario
• Traffic shaping is highly variable across users• Access link characteristics affect performance• Modem buffers induce high latency
8
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Interpreting Throughput Results
Different techniques measure different aspects of throughput
9
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Results: Overview
• Throughput measurement technique depends on usage scenario
• Traffic shaping is highly variable across users• Access link characteristics affect performance• Modem buffers induce high latency
10
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Traffic Shaping: PowerBoost
• Cable companies advertise “PowerBoost”– Short bursts of high bandwidth
• Non-existent in DSL
11
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Traffic Shaping Varies Across Users
Short-term throughput significantly different from sustainable throughput
12
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Results: Overview
• Throughput measurement technique depends on usage scenario
• Traffic shaping is highly variable across users• Access link characteristics affect performance• Modem buffers induce high latency
13
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Latency Measurements
14
Parameter What it captures
End-to-end Latency to nearby server
Last-mile Latency to edge of ISP network
Under Load Buffer delays due to cross traffic
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
DSL last-mile latencies can be very high
Impact of Last-mile on Latency
15
Cable ISPs DSL ISPs
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
DSL Interleaving Affects Latency
9 8 7 6 5 4 3 2 1
Fastpath is susceptible to bursty loss
9 6 3 8 5 2 7 4 1
Fastpath sends data in order, can recover from single losses
Interleaving sends data out-of-order, can recover from bursty loss
Fastpath vs. Interleaved last-mile data path
16
Extra Delay
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Example: Latency-Throughput Tradeoff
17
User 2: high latency, low loss
User 1: low latency, high loss
Thro
ughp
ut (M
bits
/s)
Latency (ms)10
2.5
2.2
30
Both users have same service plan
Interleaving decreases loss, increases latency, improves throughput
Interleaving creates a trade-off between latency and throughput
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Results: Overview
• Throughput measurement technique depends on usage scenario
• Traffic shaping is highly variable across users• Access link characteristics affect performance• Modem buffers induce high latency
18
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Modem Buffers are Too Large
Service plans can interact badly with modem buffers
19
10 seconds!
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Conclusion
• The gateway provides unique insight into home network
• Throughput measurements are affected by measurement technique, shaping
• Latency is affected by last-mile, buffering• ISPs generally deliver consistent throughput,
with some time-of-day effects• How to mitigate buffering effects
20
Broadband Internet Performance: A View from the Gateway ACM SIGCOMM, Toronto., August 2011
Future Work
• BISmark node deployment across the world– Currently nodes in US, South Africa
• Performance monitoring from the edge– Effect of peering on performance– IPv6 performance– Effect of CDN location, traffic engineering on
application performance• Measure the home network
21