|
|
|
|
|
|
|
|
|
|
|
|
| home | help | back | first | fref | pref | prev | next | nref | lref | last | post |
From: Masataka Ohta <mohta@necom830.hpcl.titech.ac.jp> Message-Id: <200105261133.UAA03478@necom830.hpcl.titech.ac.jp> In-Reply-To: <20010525205912.2642.cpmta@c004.sfo.cp.net> from Sean Donelan at "May 25, 2001 01:59:12 pm" To: Sean Donelan <sean@donelan.com> Date: Sat, 26 May 2001 20:33:35 +0859 () Cc: nanog@merit.edu Errors-To: owner-nanog-outgoing@merit.edu Sean; > Gamers are very away of network latency, and can produce > fairly decent reports describing what is important to > them. > > Here is a nice report I saw on slashdot about the limits > of acceptable latency for Quake gamers. > > http://members.home.net/garmitage/things/quake3-latency-051701.html It may be interpreted as an acceptable packet loss ratio at which observed latency is common. Or, it may be just that at the queue length for the latency, RED algorithm of common routers start dropping significant amount of packets. Masataka Ohta PS FYI, if you ignore diffserve hype, it is easy to reduce latency below 150msec with scalable per-flow resource reservation, though it will be charged proportional to the length of reservation.
|
|
|
|
|
|
|
|
|
|
|
|
| home | help | back | first | fref | pref | prev | next | nref | lref | last | post |