ISP Throttling?

I have high-speed cable, and when I visit performance testing web sites,
which use file transfers (over HTTP I believe), things look very good.

And Second Life used to perform noticably better than my previous
network service...for the first few weeks. But for the last week or so,
SL lag has been very bad for me. Doesn't matter much where I am.
Textures take forever to load, movement is degraded, etc.

I am wondering if Cox is throttling my connection to SL (but maybe
not to other services).

What would be a good way to tell?

Or is it possible that there are major network problems at LL's end persisting for more than a week?

SL performs well at 0.5 Mbps, and in fact some people insist it's best to limit it to that. Your cable service is probably at least 6 Mbps. They'd have to be throttling SL quite a bit. That's probably not what's happening.

This and LL server performance issues. I think SL has been slow from most folks the last few weeks.

LL has been having some performance issues in the last couple of weeks, but I personally think these are more server-related and not network issues.

Cox Cable was one of the first service providers to use traffic shaping to limit bandwidth, and they still do so, although enforcement varies across the various markets they sell into.

DSLReports, as usual, has relevant info: http://www.dslreports.com/shownews/Cox-Also-Disrupting-P2P-Traffic-89481 According to them, Cox is not 'capping' bandwidth, they are inserting poison packets to force certain applications (or types of applications) to back down. Another common technique is to shape traffic for certain applications down to a small percentage of what they need, which in many cases forces the application to back down, without the user being able to detect what is happening. Either of these techniques will throttle a single apps performance, even in cases where the individual user has loads of available bandwidth - the ISP's goal is to limit that type of traffic on their backbone, not limit it on an individual user's line.

p.s. How to tell ... well, you MIGHT be able to see the signature of traffic shaping on your Task Manager network performance graph, if you watch it carefully when you start SL. You could also use an analyser like WireShark. Or you could call Cox, if your confrontation skills are good enough to make it through 1st and 2nd level tech support, and reach someone who knows something.
.

Try rebooting your modem and router. Unplug the power cord for a minute, and then plug em back in. I do this at least once a week. Often performance is much better. I believe its my own router firewall that gets in the way.

maybe the current swithover to the new LL netwerk is what is causing this, afterall at the moment sl is not made up of just one VPN but a mixture of a VPN and a true private network, hopefully it will get better once the move to their priate netwerk is completed.

Grab ISP by the neck. Shake.

It's more than possible!

If your concerned about ISPs and the possibility they may be putting your internet on the back burner in favor of someone else, then this is THE site to bookmark and check often...
http://www.measurementlab.net/measurement-lab-tools

EDIT: So you know, this is the site that was created in response to Comcast and what they were doing with bittorrent traffic.

thanks for that link my ISP and system failed badly.

Test conditions
Tester: xxxxxxxxxx (xxxxxxxx) [?]
Target: On-05-0129.dsl.xxxxxxxx.net (xx.xxx.xxx.xxxxx) [?]
Logfile base name: On-05-0129.dsl.xxxxxxxxxxxx.net:2009-04-04-11:50:42 [?]
This report is based on a 10 Mb/s target application data rate [?]
This report is based on a 8001 ms Round-Trip-Time (RTT) to the target application [?]
The Round Trip Time for this path section is 60.000000 ms.
The Maximum Segment Size for this path section is 1460 Bytes. [?]

Target host TCP configuration test: Fail! [?]
Warning: TCP connection is not using RFC1323 timestamps. [?]


Critical Failure: Received window scale is 0, it should be 9. [?]
The maximum receiver window (17k) is too small for this application (and/or some tests). [?]
Diagnosis: The target (client) is not properly configured. [?]
> See TCP tuning instructions at http://www.psc.edu/networking/projects/tcptune/ [?]
Path measurements [?]

Data rate test: Fail! [?]
The maximum data rate was 1.658558 Mb/s. [?]
This is below the target rate (10.000000 Mb/s). [?]
This test did not complete due to other problems with the path, target or tester.
> Correct other problems first, and then rerun this test. [?]

Loss rate test: Fail! [?]
Loss rate measurement based on insufficient data and may be inaccurate. [?]
The data rate was not high enough to accumulate sufficient loss statistics in a reasonable amount of time. [?]

> Correct data rate problems and re-test the path. [?]

Fail: loss event rate: 0.007158% (13971 packets between loss events). [?]

Diagnosis: there is too much background (non-congested) packet loss. [?]

The events averaged 1.333333 losses each, for a total loss rate of 0.009543%. [?]

FYI: To get 10 Mb/s with a 1460 byte MSS on a 8001 ms path the total end-to-end loss budget is 0.000001% (95764991 packets between losses). [?]

> Locate the excess packet loss in this section of the path. [?]
> Localize all path problems by testing progressively smaller sections of the full path. [?]


Tester validation: Pass! [?]
No internal tester problems were detected.
///////////////////////////////////////////////////////////////////////////

I followed up all the little (?) to links but its way beyond me on how to fix my window reciever sizes and the rest....

thanks that was MY next step anyway since the telephone company took over my cable company, my 8Mbit line has been running between ~1.3Mbit and 12Kbit (that's slower than dial-up). tech says signal is good, my upstream is always good, but the down stream, in addition to , to the poor speeds, has seen packet loss of over 50%.... furious is an understatement. of course the line techs have no clue why it's happening, but I have some serious suspicions.

@ciera:
DslReports (akaBroadbandReports) .com has great tools for ajusting TCP recieve window on your end (though this helps nothing if the isp chokes theirs). it will take some reading though.

@general
if your isp advertised speed is > 3Mbit the java tests are more accurate (plus I see a trend in the flash based ones of recording the high water mark, even it that was only barely available compared to lower speeds) the java test is susceptible to hiccups at higher speeds, but you can see that happening as a pause in the progress bar before resuming... usually you want to retest those.

Ooo, nasty! I doubt any of the problems that test found are causing you any real trouble. If you're just trying to test your actual broadband performance, try something easier. I use this one:

http://www.dslreports.com/speedtest

I generally use the Flash-based test. Pick a test server that's in either California or Texas.

However, if your ISP is throttling apps like bittorrent or streaming video like SL, the test won't detect that.

(p.s. the suggestion to first just try power cycling your modem (I'd leave it off for 15 minutes, so any funky ARP entries time out), is a good one.
.

well I found a lot of help in the setting of RWIN but i dont dare do it. my computer wont allow me to do a restore if i mess it up.

it looks liek from my test s i have a RWIN of 17,525 when im being told I need 134,425.. thats a huge difference. BUT I just talked to my service provider and the told me best i was ever going to get in download speeds is 1.2mps

SO its a moot point then.

As it is I Don't dare mes it up.

You don't have a good backup of your computer? You don't have installation media? I recommend you do something about that post haste.

Mine had the same exact results until I realized the default setting of 10 Mbps is a target you have to reach to pass, so I lowered it and got better results.