cc_cdg —
CDG
Congestion Control Algorithm
CAIA-Delay Gradient (CDG) is a hybrid congestion control algorithm which reacts
to both packet loss and inferred queuing delay. It attempts to operate as a
delay-based algorithm where possible, but utilises heuristics to detect
loss-based TCP cross traffic and will compete effectively as required. CDG is
therefore incrementally deployable and suitable for use on shared networks.
During delay-based operation, CDG uses a delay-gradient based probabilistic
backoff mechanism, and will also try to infer non congestion related packet
losses and avoid backing off when they occur. During loss-based operation, CDG
essentially reverts to
cc_newreno(4)-like
behaviour.
CDG switches to loss-based operation when it detects that a configurable number
of consecutive delay-based backoffs have had no measurable effect. It
periodically attempts to return to delay-based operation, but will keep
switching back to loss-based operation as required.
The algorithm exposes the following variables in the
net.inet.tcp.cc.cdg branch of the
sysctl(3) MIB:
- version
- Current algorithm/implementation version number.
- beta_delay
- Delay-based window decrease factor as a percentage (on
delay-based backoff, w = w * beta_delay / 100). Default is 70.
- beta_loss
- Loss-based window decrease factor as a percentage (on
loss-based backoff, w = w * beta_loss / 100). Default is 50.
- exp_backoff_scale
- Scaling parameter for the probabilistic exponential
backoff. Default is 2.
- smoothing_factor
- Number of samples used for moving average smoothing (0
means no smoothing). Default is 8.
- loss_compete_consec_cong
- Number of consecutive delay-gradient based congestion
episodes which will trigger loss-based CC compatibility. Default is
5.
- loss_compete_hold_backoff
- Number of consecutive delay-gradient based congestion
episodes to hold the window backoff for loss-based CC compatibility.
Default is 5.
- alpha_inc
- If non-zero, this enables an experimental mode where CDG's
window increase factor (alpha) is increased by 1 MSS every
alpha_inc RTTs during congestion
avoidance mode. (Setting alpha_inc to 1
results in the most aggressive growth of the window increase factor over
time. Use higher alpha_inc values for
slower growth.) Default is 0.
cc_chd(4),
cc_cubic(4),
cc_hd(4),
cc_htcp(4),
cc_newreno(4),
cc_vegas(4),
h_ertt(4),
mod_cc(4),
tcp(4),
khelp(9),
mod_cc(9)
D. A. Hayes and
G. Armitage, Revisiting TCP
Congestion Control using Delay Gradients, Networking
2011 Proceedings, Part II, 328-341,
May 2011.
N. Khademi and
G. Armitage, Minimising RTT across
homogeneous 802.11 WLANs with CAIA Delay-Gradient TCP (v0.1),
CAIA Technical Report 121113A,
http://caia.swin.edu.au/reports/121113A/CAIA-TR-121113A.pdf,
November 2012.
Development and testing of this software were made possible in part by grants
from the FreeBSD Foundation and The Cisco University Research Program Fund, a
corporate advised fund of Silicon Valley Community Foundation.
The
cc_cdg congestion control module first appeared
in
FreeBSD 9.2.
The module was first released in 2011 by David Hayes whilst working on the
NewTCP research project at Swinburne University of Technology's Centre for
Advanced Internet Architectures, Melbourne, Australia. More details are
available at:
http://caia.swin.edu.au/urp/newtcp/
The
cc_cdg congestion control module was written by
David Hayes
<
[email protected]>.
This manual page was written by
Lawrence
Stewart
<
[email protected]>
and
Grenville Armitage
<
[email protected]>.
The underlying algorithm and parameter values are still a work in progress and
may not be optimal for some network scenarios.