The Power of Explicit Congestion Notification Aleksandar Kuzmanovic - - PowerPoint PPT Presentation
The Power of Explicit Congestion Notification Aleksandar Kuzmanovic - - PowerPoint PPT Presentation
The Power of Explicit Congestion Notification Aleksandar Kuzmanovic Northwestern University http://www.cs.northwestern.edu/~akuzma/ http://www.cs.northwestern.edu/~akuzma/ Motivation Recent measurements [PF01,MPF04]: 2000: 1.1% Web
2
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Motivation
Recent measurements [PF01,MPF04]:
– 2000: 1.1% Web servers support ECN – 2004: the percent increased to 2.1% – Not a single packet was marked in the network
2100 2200 2300 2400 year 2000 50% 100% ECN deployment ye year ar 239 2396 (Sigc Sigcomm
- mm # 41
# 411) 1)
3
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Background
Server Client Marker/ Dropper FCFS scheduler AQM Router
Active Queue Management (AQM):
– Simultaneously achieves high throughput and low average delay – AQM algorithms can mark (instead of drop) packets – The router and both endpoints have to be ECN-enabled
4
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Negotiating ECN Capabilities (I)
Client initiates ECN-capable communication by setting appropriate bits in the TCP SYN packet’s TCP header TCP header
Server Client TCP SYN
5
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Negotiating ECN Capabilities (II)
An ECN-capable server replies by setting appropriate bits in the SYN ACK packet’s TCP header TCP header
Server Client TCP SYN SYN ACK
Once the SYN ACK packet arrives, ECN negotiation is completed
6
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Barriers to Adoption of ECN
Server Client TCP SYN reset router "Broken" firewall
“Broken” firewalls and load balancers incorrectly reset TCP flows attempting to negotiate ECN
– The problem addressed in RFC 3360
Consequences are devastating New incentives?
7
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
ECN and TCP’s Control Packets
Server Client TCP SYN SYN ACK HTTP REQ
TCP SYN and SYN ACK packets are dropped during congestion Can significantly reduce end-to-end performance
– RTO = 3 sec (+6 sec, +12 sec, etc.)
Marking SYN packets?
8
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Marking TCP SYN Packets?
Server Client TCP SYN
TCP SYN packets:
– Security problems
SYN ACK packets:
– No security obstacles – More relevant
- Congestion likely to happen
from servers to clients
9
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Marking SYN ACK Packets?
Server Client TCP SYN SYN ACK
TCP SYN packets:
– Security problems
SYN ACK packets (ECN+):
– No security obstacles – More relevant
- Congestion likely to happen
from servers to clients
10
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Deployment Requirements
Security
– No novel security holes
Performance improvements
– Necessary to provide incentives to all involved parties
Incremental deployability
– What level of deployment is needed to achieve the above improvements? – What happens to those who do not apply the change?
11
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Simulation Scenario
Client pool Server pool 1 Gbps 1 Gbps 100/622/1,000 Mbps
Server Server Server Client Client Client
requests responses AQM
90% objects downloaded in less than 0.5 sec
Light and persistent congestion from servers to clients Web and general traffic mixes AQM algorithms: Random Early Detection (RED) (others in the paper)
12
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Outdated Implementation
min_th max_th
Average Queue Length
max_p 100%
Drop/mark rate
RED (1993)
– “This notification can consist of dropping or marking a packet.”
RFC 3168 (2001)
– Guidelines for setting ECN with RED
Older RED versions still present (e.g., Linux)
RED’s dropping/marking rate as a function of the queue length
13
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Dropping RED
min_th max_th
Average Queue Length
max_p 100%
Drop/mark rate
- perating
point
Reduced performance due to congestion
14
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Add ECN
min_th max_th
Average Queue Length
max_p 100%
Drop/mark rate
- perating
point
All SYN packets are dropped Outdated implementation can cause drastic performance degradations
15
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Add ECN+
min_th max_th
Average Queue Length
max_p 100%
Drop/mark rate
- perating
point
ECN+ systematically improves throughput and response times of all investigated AQM schemes SYN ACK packets are NOT dropped
16
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Incremental Deployability
Scenario
Client pool Server pool
Server Server Server Client Client Client
ECN+ at servers ECN at routers x% clients: ECN (100-x)%: no ECN
17
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
5% Deployment
Instant gains for ECN-enabled clients 95% no ECN 5% ECN
18
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
50% Deployment
50% ECN 50% no ECN Gradual degradation for clients not applying ECN
19
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
95% Deployment
95% ECN 5% no ECN Performance necessarily degraded
20
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
Testbed Experiments
Server pool 100 Mbps 10 Mbps
Server Server Server Client
requests responses (15 Mbps) router ECN no ECN ECN ECN+
21
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN
ECN and Flash Crowds
RED, no ECN RED, ECN RED, ECN+ Average Response Time Throughput (% of capacity)
26 sec 4.5 sec 0.5 sec 44% 56% 99%
Reasonable performance despite huge congestion
22
- A. Kuzmanovic
- A. Kuzmanovic
The Power of ECN The Power of ECN