[or-cvs] r18674: {projects} Proposal to reduce circuit window size (projects/performance)

[sjm217 at seul.org]

Author: sjm217
Date: 2009-02-22 09:50:37 -0500 (Sun, 22 Feb 2009)
New Revision: 18674

Modified:
   projects/performance/performance.bib
   projects/performance/performance.tex
Log:
Proposal to reduce circuit window size

Modified: projects/performance/performance.bib
===================================================================
--- projects/performance/performance.bib	2009-02-22 10:23:34 UTC (rev 18673)
+++ projects/performance/performance.bib	2009-02-22 14:50:37 UTC (rev 18674)
@@ -59,6 +59,15 @@
   note = 	 {\url{http://archives.seul.org/or/dev/Dec-2008/msg00005.html}},
 }
 
+ at Misc{circuit-window,
+  author = 	 {Csaba Kiraly},
+  title = 	 {Effect of {Tor} window size on performance},
+  month = 	 {February},
+  howpublished = {Email to or-dev at freehaven.net},
+  year = 	 {2009},
+  note = 	 {\url{http://archives.seul.org/or/dev/Feb-2009/msg00000.html}},
+}
+
 @mastersthesis{reardon-thesis,
   title = {Improving {Tor} using a {TCP}-over-{DTLS} Tunnel}, 
   author = {Reardon, Joel}, 

Modified: projects/performance/performance.tex
===================================================================
--- projects/performance/performance.tex	2009-02-22 10:23:34 UTC (rev 18673)
+++ projects/performance/performance.tex	2009-02-22 14:50:37 UTC (rev 18674)
@@ -240,6 +240,18 @@
 There are few options available, as TCP is significantly more popular.
 Voice over IP is one fruitful area, as these require low latency and hence UDP is common, but further investigation is needed.
 
+\section{Changing circuit window size}
+
+Tor maintains a per-circuit maximum of unacknowledged cells (\texttt{CIRCWINDOW}).
+If this value is exceeded, it is assumed that the circuit has become congested, and so the originator stops sending.
+Kiraly proposed~\cite{circuit-window,tor-l3-approach} that reducing this window size would substantially decrease latency (although not to the same extent as moving to a unreliable link protocol), while not affecting throughput.
+This reduction would improve user experience, and have the added benefit of reducing memory usage on Tor nodes.
+
+More investigation is needed on precisely what should be the new value for the circuit window, and whether it should vary.
+Out of 200, 1\,000 (current value in Tor) and 5\,000, the optimum was 200 for all levels of packet loss.
+However this was only evaluated for a fixed network latency and node bandwidth.
+Therefore, a different optimum may exist for networks with different characteristics.
+
 \section{Tor server advocacy}
 
 Encouraging more volunteers to run Tor servers, and existing volunteers to keep their servers running, would increase network capacity and hence performance.