[Bloat] A hard ceiling for the size of a drop-tail queue

[Jonathan Morton]

On 11 Mar, 2011, at 3:15 am, Jonathan Morton wrote:

> So how many bytes is a 2-second queue on a wireless network?  It can be no more than the *minimum* link speed (in bytes/sec) divided by the *maximum* number of nodes.  This allows a 100% overhead for framing and congestion avoidance.  For 802.11 the minimum link speed is 1Mbps = 100KB/s, so with 100 nodes (not a very large conference) the buffer should be no more than 1KB - which is less than one full packet.  And since there are only about 3 non-overlapping channels in the 2.4GHz band, then even if you spread those 100 nodes over 3 base stations, you only get 3KB = 2 packets of buffer space.  (And this includes the base station!)
> 
> So I'm surprised that these conference networks ever function at all.

Coming back to this topic for a moment, I have a radical solution to the "wifi congestion collapse" problem:

If a packet has been in the send queue for more than 1 second, drop it - regardless of whether it has been sent or not.

Why?  Because if the application and the user are still interested in getting the traffic through, they will retry.  If not, dropping stale packets will stop thee retries (which will happen anyway) from clogging up the network.  The first TCP-SYN retry will be after 3 seconds, so dropping the first try after 1 second gives back 2/3rds of the airtime and prevents the queue growing.

Dave Täht gave me something to listen to, and one of the things that was mentioned was that the aggregation aims for 4ms of airtime at once.  This means that there is time for about 250 aggregate-frames per second.  This, to me, puts the sustainable number of nodes at about 100 as I suspected.  But remember that at the minimum rate of 1Mbps, 4ms is only 500 bytes - enough for TCP SYN or DNS to work, but only one-third of a full Ethernet frame.

 - Jonathan