[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Tcp vegas

To: Rajib Ghosh <[email protected]>
Subject: Re: Tcp vegas
From: John Heidemann <[email protected]>
Date: Fri, 13 Feb 1998 14:19:28 -0800
Cc: ns users <[email protected]>
In-reply-to: <[email protected]>

On Fri, 13 Feb 1998 15:42:58 CST, Rajib Ghosh wrote: 
>I am a graduate student working on TCP congestion control. I have been using NS for my simulations for a while now. However the implementation of TCP vegas seems to be incorrect in ns-2.1b1. i tried ns2.1b2 but that seems to be incorrect too. can somebody throw some light on this?

Can you be more specific than ``seems to be incorrect''?

To my knowledge, VINT folks have only looked at Vegas slow-start
behavior in detail.  2.1b1 Vegas slow-start behavior has serious
broken, but it should be fixed in 2.1b2.  The 2.1b2 Vegas should match
what ns-1's Vegas did (according to the test suite) and appears to
exhibit reasonable Vegas-style slow-start behavior.

Our test suite coverage is far from complete, though.  Maybe you're
looking at something other, less-debugged Vegas behavior?

   -John Heidemann

References:
- Tcp vegas
  - From: Rajib Ghosh <[email protected]>

Prev by Date: Tcp vegas
Next by Date: a glitch in NS-ALL-IN-ONE
Prev by thread: Tcp vegas
Next by thread: simulating RLM in nsv2
Index(es):
- Date
- Thread