Here are some stats from an RMON probe on one of our broadband consumer
Internet access networks. It's sitting between our access network and our
Internet transit connections, so it sees every packet coming from or going
to the subscribers. This represents about 7TB worth of data over the last
few months.
Packet Size Distribution
All numbers are percentages
Downstream Upstream
---------- --------
0 - 64 14.68 58.49
65 - 127 13.87 29.73
128 - 255 7.25 1.72
256 - 511 6.44 3.98
512 - 1023 13.59 3.37
1024 - 1518 44.17 2.70
It's interesting to note the disparity between the upstream and downstream
distributions. I was also surprised to see that half of the packets on the
downstream are significantly larger than 576. I just wish I had a bucket
boundary at 576+18. Still, there's no ambiguity about the 44% of packets
that are (probably) at Ethernet MTU.
Steve Dispensa
Sr. Network Architect
Sprint Broadband Wireless Group
-----Original Message-----
From: Stuart Cheshire [mailto:cheshire(_at_)apple(_dot_)com]
Sent: Thursday, May 25, 2000 9:10 PM
To: ietf(_at_)ietf(_dot_)org
Subject: Re: Average Ethernet packet length
There are no good, current studies on LAN behavior that I've seen.
There have been a number of papers on WAN behavior. The usual result
of those is that ~40-50% of packets are about 40-44 bytes, but most of
the bytes are carried by packets of ~500-576 or 1500 bytes.
--Steve Bellovin
In some traces I did for my PhD work about three years ago, I found that
51% of the packets were 40 or 41 bytes long (i.e. mostly TCP acks or
one-byte TCP payloads). Only 15% of the packets were maximum-sized. The
average packet size was 273 bytes. This workload was probably heavier on
telnet than today's networks; however even when doing bulk file transfer,
one packet in three is still an ack.
Stuart Cheshire <cheshire(_at_)apple(_dot_)com>
* Wizard Without Portfolio, Apple Computer