Re: Emulating ADSL bandwidth shaping
--- cboyd@gizmopartners.com wrote: From: Chris Boyd <cboyd@gizmopartners.com> On May 4, 2010, at 7:27 AM, Marshall Eubanks wrote:
I am not sure what the point is in mixing in speed of light latency. If your "typical sites" are, say, Indian cricket blogs, you will typically have a high latency from the US. What does that tell you about your DSL or Cable system, except that it is somewhat removed from India ?
: Most of the ADSL installations I've seen in SBC 13 state area : had interleaving turned on, which significantly increases latency. : I suspect that's why many cable MSOs in the same territory have : "cable is better for gaming" marketing campaigns running all the time. : : So the latency you see on an ADSL line is dependent on how the : carrier set up the DSLAM. In the system I work on, which is in a state with many rural areas and an OSP that goes through a lot of very wet and very windy areas, has "Interleaved" turned on to correct errors. This adds ~25msec between the CPE and the nearest router. Sometimes folks ask for it to be changed to "Fast". We explain that errors may cause resyncs to happen and then make the change if the customer still wants it. That takes the latency from ~25msec to ~3-4msec. So use both when doing your modeling. I don't know how to model the error rate, though, as that would be dependent on the quality of the OSP you're modeling... scott ps. if you're so fast in gaming that a difference of 1/5 of a second makes a difference you're good... ;-)
On Tue, 4 May 2010, Scott Weeks wrote:
"Interleaved" turned on to correct errors. This adds ~25msec between the CPE and the nearest router. Sometimes folks ask for it to be changed to "Fast". We explain that errors may cause resyncs to happen and then make the change if the customer still wants it. That takes the
Basically "interleaved" is a way "smearing" bits over time, so FEC can do its job when there is a 1ms line hit that causes a lot of errors during that 1ms. Turning off interleaving (changing to "fast") usually means your "errored seconds" counter will increase over time because of lack of error correction, and your customer experiences packet loss intermittently. Never seen it cause resyncs, but that might happen, dunno. It is usually settable to 1,4,16 ms in each direction, so 25ms is not a "law of nature", it's just the default 16/4 ms downstream/upstream interleaving and then the added coding delay. -- Mikael Abrahamsson email: swmike@swm.pp.se
participants (2)
-
Mikael Abrahamsson
-
Scott Weeks