Performance Andre Ryll, B.Eng. Content TCP Basics revisited Facts - - PowerPoint PPT Presentation

Lehrstuhl Netzarchitekturen und Netzdienste

Institut für Informatik Technische Universität München

Measuring TCP Tail Loss Probe Performance

Andre Ryll, B.Eng.

TCP Tail Loss Probe

2

Content

 TCP Basics revisited

• Facts
• TCP Header

 TCP Loss Recovery

• Duplicate ACKs
• Selective ACKs
• Retransmission Timeout

 Tail Loss Probe  Test Setup  Measurements  Results  Conclusion

TCP Tail Loss Probe

3

TCP Basics – What you should know

 ISO OSI Layer 4 (Transport Protocol)  Connection oriented  Reliable (retransmissions)  Ordered (sequence numbers)  Error-Checked (checksum)  Stream of octets  Counterpart: UDP (message oriented, unreliable, out-of-order)

TCP Tail Loss Probe

4

 SYN flag

Synchronize sequence number, set once during connection setup

 FIN flag

No more data from sender

 ACK flag

ACK field is valid

 acknowledgment number

Next expected sequence number

 sequence number

Octet sequence number

TCP Basics – Header

TCP Tail Loss Probe

5

TCP Example Flow (Tail)

TCP Tail Loss Probe

6

TCP Basics

How does TCP recover from loss?

TCP Tail Loss Probe

7

TCP Loss – Duplicate ACK (DACK)

TCP Tail Loss Probe

8

TCP Loss – Selective ACK (SACK)

TCP Tail Loss Probe

9

TCP Basics

What if the tail of the message is lost?

TCP Tail Loss Probe

10

TCP Loss – Retransmission Timeout (RTO)

TCP Tail Loss Probe

11

Tail Loss Probe (TLP)

TCP Tail Loss Probe

12

Tail Loss Probe (TLP)

 A: ACKed segment  L: Lost segment  S: SACKed segment

losses After TLP Repair AAAL AAAA TLP Probe AALL AALS Early Retransmit (ER) ALLL ALLS Early Retransmit (ER) LLLL LLLS Forward ACK >=5 L ..LS Forward ACK

TCP Tail Loss Probe

13

Test Setup

Mininet Virtual Network

 Lightweight and fast network

setup with kernel support

 nginx webserver  lynx webbrowser with –dump

• ption

 iptables for packet filtering

TCP Tail Loss Probe

14

Test Setup

Mininet Virtual Network

 Lightweight and fast network

setup with kernel support

 nginx webserver  lynx webbrowser with –dump

• ption

 iptables for packet filtering

H1 configuration

 Inbound traffic passes to nginx  Outbound  non-TCP pass  TCP  NFQUEUE  C/C++ user-space application

„tcpfilter“ tracks HTTP data

 Last segments are dropped

TCP Tail Loss Probe

15

Measurements

 Transfer size: 100 segments (approx. 144kB)  Varying tail loss: 0 – 20 segments  Varying network conditions  sysctl interface to change TCP parameters  Measured:

• Time to first retransmitted packet: trecover
• Total transfer time: ttotal

Option plain ER TLP tcp_early_retrans 2 3 tcp_fack 1 1 tcp_sack 1 1 Type Bandwidth RTT high-speed 100MBit/s 10ms mobile 7.2MBit/s 100ms satellite 1MBit/s 800ms

TCP Tail Loss Probe

16

Results – Time to retransmit

high-speed network

Legend: plain ER TLP

TCP Tail Loss Probe

17

Results – Total transfer time

high-speed mobile satellite

Legend: plain ER TLP

TCP Tail Loss Probe

18

Results – Numbers!

high-speed Plain ER TLP ttotal 0.3595 +0.6% 0.3574 0.2214

• 38%

trecover 0.2396 +1.3% 0.2365 0.0447

• 81%

satellite Plain ER TLP ttotal 8.2145 -0.1% 8.2248 8.1948

• 0.4%

trecover 2.4792 +0.3% 2.4720 2.4331

• 1.6%

mobile Plain ER TLP ttotal 1.2091 -1.7% 1.2300 1.0944

• 11%

trecover 0.4392 +7.8% 0.4073 0.3310

• 19%

(5 lost packets)

TCP Tail Loss Probe

19

Conclusion

 TLP is an improvement in all tested cases  Time to first retransmission reduced by up to 81%  Total transfer time reduced by up to 38%  My measurements show:

• No benefit for single lost segment
• Little benefit in high-delay networks
• Although it would be very important in those networks

TCP Tail Loss Probe

20

QUESTIONS?

Thank you!