Considerations of SCTP Retransmission Delays for Thin Streams Jon Pedersen

advertisement
LCN 2006: 31st IEEE Conference on Local Computer Networks, Tampa, FL, USA, November 2006
Considerations of
SCTP Retransmission Delays
for Thin Streams
Jon Pedersen1, Carsten Griwodz1,2 & Pål Halvorsen1,2
1Department
of Informatics, University of Oslo, Norway
2Simula Research Laboratory, Norway
{jonped, griff, paalh}@ifi.uio.no
Overview
 Latency problems for thin streams
 SCTP as an alternative to TCP
 Experiments
 New experiments
 Conclusions
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Thins Streams
 Transport protocols being developed for throughput-bound applications
 BUT, there exist several low-rate, time-dependent applications
 Anarchy Online MMORPG Case Study





average delay:
max delay:
packets per second:
average packet size:
average bandwidth requirement:
LCN 2006, Tampa, FL, USA, November 2006
~250 ms
67 seconds (6 retransmissions)
<4
(less then one per RTT)
~120 bytes
~4 Kbps
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Stream Control Transmission Protocol
sender
 SCTP should support signaling







“require response between 500 – 1200 ms” … or
“initiation of error procedures” [rfc 2719]
Network
SACK
(re)transmission queue

acknowledged error-free transfers
data fragmentation according to MTU
packet boundary maintenance
sequenced delivery within multiple streams
bundling
partial reliability
…
suppose to address low latencies
receiver
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Test Set Up
 Linux 2.6.15 with lksctp
 100 bytes packets
 4 packets per second  3.2 Kbps
SCTP
LCN 2006, Tampa, FL, USA, November 2006
Network emulated using netem
• dropp
• delays
(RTTs: 0, 100, 200, 400 ms)
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Results: lksctp for Thins Streams
 Even worse than TCP!!!
 Why these high delays?
 Two ways of triggering retransmissions of a lost chunk…
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Retransmission by Time-Out
sender
 Timeout is dependent on

minRTO = 1000 ms

estimated RTT based on SACKs

BUT SACKs are delayed
o
o
one ACK for two packets or
200 ms timer
 influences estimated RTT,
for thin streams
retransmissionespecially
of packet with
green chunks due to timeout
 RTO value grows
Network
SACKSACK
(re)transmission queue
receiver
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Retransmission by Fast Retransmit
sender
receiver
4 SACKs needed for fast retransmit
+ thin streams
= “all” retransmissions due to timeouts
no
no
no
no
SACK
SACK
SACK
SACK
Network
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Enhancement: Removal of Exponential Backoff
time in RTTS
sender
receiver
8
6
4
(re)transmission queue
2
1
2
3
4
retransmission number
retransmission of
ENHANCEMENT:
green packet
remove exponential backoff
due to timeout
LCN 2006, Tampa, FL, USA, November 2006
Network
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Enhancement: Fast Retransmit Bundling
sender
receiver
ENHANCEMENT:
piggyback all chunks in retransmission queue
blue packet is NOT piggybacked when dupACKs
(but would be if due to timeout)
Network
no
no
no
no
SACK
SACK
SACK
SACK
retransmission queue
retransmission of green packet (chunks) due to dupACKs
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Enhancements
 Modified retransmission timer
 removal of exponential backoff
 minRTO = 200 ms (as in TCP)
 Modified retransmission bundling
 always allow aggressive bundling for fast retransmit
 Modified fast retransmit
 tested fast retransmit after 1 SACK
 Thin stream detection
 fewer packets in flight to trigger a fast retransmit
 added tracking of outstanding packets
 less than 4 in flight = thin stream
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Enhancement: Results (200 ms)
 Considerable reduction in
average and maximum
latencies
 Increase in number of fast
retransmissions compared to
timeouts
 Increase in number of
retransmissions
original SCTP
Timeout
Fast retransmit
Total
LCN 2006, Tampa, FL, USA, November 2006
reduced minRTO &
fast retransmit
modified timer restart
no SACK delay
266
197
331
633
35
284
288
1
301
481
619
634
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
New lksctp versions & New Test Set Up
 New lksctp versions has been developed

lksctp in 2.6.16 (2.5.72-0.7.1)



only one retransmission due to fast
retransmit, next timeout
only 3 SACKs required for fast retransmits
lksctp in 2.6.17 has no major changes
for our scenario
 New tests
 100 B packets
 RTTs:
0, 50, 100, 150, 200, 250 ms
 Packet inter-arrival times:
50, 100, 150, 200, 250 ms
 Dynamic thin stream detection

WEB
SCTP
Many web-connections generating
cross traffic (and thus losses)
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Results: New lksctp
 Still high average and worst case latencies
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Results: Fast Retransmission Modification
Fast retransmit modification – 1 SACK
 Reduction in maximum and average latency
 As expected a large increase in fast retransmit
 An increase in spurious retransmissions
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Results: Removed Exponential Backoff
Removed exponential backoff
 Reduction in maximum and average latency
 An increase in spurious retransmissions
 Retransmission aggressiveness does not really pose a congestion threat
since the amount of data waiting to be sent is always less than the minimum
transmission window
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Results: Reduced Minimum Time-out
Reduced minRTO
 Faster timeouts
 Reduction in maximum and average latency
 An increase in spurious retransmissions
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Results: All Modifications Combined
All thin stream modifications
 A further reduction in maximum and average latency
 As expected an increase in fast retransmit
 An increase in spurious retransmissions
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Conclusions
 Based on SCTP description we expected (hoped for) reduced
latencies compared to TCP
 Enhancements like
 reduced minRTO
 removal of exponential backoff
 removal of delayed SACKs
 …
reduce latencies for thin streams
 The enhancements increase the number of spurious
retransmissions, but maybe not important for thin streams!!??
LCN 2006, Tampa, FL, USA, November 2006
2006 Jon Pedersen, Carsten Griwodz & Pål Halvorsen
Download