RT31P2 and QoS = My Results, Can People Test my Formula?

Posts: 35

Finally after many hours of experimentation I have my config in a reasonable state.

My Criteria:
1. Maximum Download Rate
2. Maximum Upload Rate
3. Voice Priority.

Setup
CBL Modem (new RCA from Comcast) -> RT31P2 (DHCP, QoS Upstream Rate=1400, see firmware in my sig) -> Internal Lan

Vonage Bandwidth Saver: 90

Comcast users
a. get rid of your old cable modem,.. if you have the 3COM (aka "Sharkfin"), or the Toshiba 1100 (there is a newer one which is supposedly okay.
b. confirm your bandwidth rates. In my case I am 6000/768.

So how did I get to the QoS Upstream Rate of 1400.

My Tests (new test results after getting rid of sharkfin junk; now using RCA):

Test 1:
Set QoS to default 256
Place call and ran bandwidth test.
While in call: 5400/117

So this is quite alarming, going from 768 to 117. Totally unncessary for a phone requirement of 90*2 = 180.

Test 2:
Set QoS to 200 (thinking if the phone needed 180, 200 should do).
While in call: 5400/83

Then I got thinking:
117/256 = 46%
and
83/200 = 42%

Now even when I was running 3000/256 for my Comcast connection, I was getting similar upload rates: 114/117/117 at QoS of default 256.

So here is my guess: The RT31P2 takes the QoS upstream rate and reserves approximately 60% for voice traffic and the remaining 40% for the user.

Taking this into consideration, I made adjustments. 768-200 = 568 (target upload rate leaving enough for voice). Doing some simple algebra: 0.4 * x = 568 results in x = 568 / 0.4 = 1420.

Rounded off my number to 1400 in favor of voice and repeated the tests:

Test 3:
Set QoS to 1400.
While in call: 5400/486

This is considerably better than before and leave more than enough bandwidth for the phone to operate successfully without bringing my upload to its knees. This also takes into consideration that perhaps you will not always hit the maximum 6000/768.

Now bear in mind.. I haven't tested with this value that much... but to see the upload rate become reasonable and in the initial test no loss in quality whatsoever leads me to believe that the RT31P2 allocates a fixed 40% of the Upstream Rate for the user. (please note my firmware/voice version). I can't say this will be the solution for everyone, but this is the only thing in a long time that has allowed me to exceed 117 without turning off QoS.

Can others try this and let me know if it solves their problems with QoS.

*** Note, QOS and QoS are not the same thing. Please be careful when using the terms.

Posts: 41 · Full Forum Member Joined: Jan 19, 2005 Posts: 41

Interesting...

I just ran a few tests. I have 4Mb down and 384kb up Comcast. I have had my QoS set to 256 since I got Vonage 2 months ago.

Here are the results of my testing, all done within a 10 minute span.

QoS 256 = 117kbps upload
QoS 350 = 124kbps upload
QoS 375 = 123kbps upload
QoS 400 = 232kbps upload
QoS 460 = 242kbps upload

460 is the result of me using your calculation of 384 - 200 / 0.4

I would have expected a lower upload than 242 with 460, but what I don't understand is why the change from 256 through 375 results in a negligible upload change, but going to 400 jumps the upload over 100kb.

I ran each speed test twice, and after going from 375 to 400 I went back to 375 and back to 400 again to confirm that the upload stayed +100. It did.

What does this mean, and what do you think the best setting would be for my QoS?

Posts: 35 · Posted: Thu Mar 10, 2005 12:14 pm Post subject:

Thats interesting indeed.

I dont have an answer to that. My testing went lower, I didn't test higher except going straight to the highest value. And the expected level being higher is odd too. Mine was lower.

I can try this at home later tonight and see if I have a similar experience. If in the meantime, there are others that can repeat this test, that would be of help.

For the time being, I would use using the 400 setting. It isn't clear whether Vonage requires 90kbps for all functionality, or if in a 3-way call you need 2*90. Test it out, if your callers are getting choppy conversations, revert to 256.

The difference in 100kbps for a small change in QoS suggests their might be more than a constant percentage. What boggles me is that no one has a single correct answer! Someone's gotta know. Vonage people dont know either, at least from the number of times I've called. Anyone have access to a technical level person at Vonage?

a. What is your network setup?
b. Might be better trying this test off peak - to minimize server load issues at speakeasy - they get busy. My testing was pretty late, 12am
c. What cable modem do you have?
d. Do you have the same firmware versions (see my signature)?

Posts: 41 · Full Forum Member Joined: Jan 19, 2005 Posts: 41

I have a Motorola Surfboard SB4100 cable modem. The connections go Modem --> RT31P2 --> Linksys WRT54G. My computer is hardwired into the RT31P2. I still have the 1.26.01 firmware.

And in my previous post I stated that the change stayed at +100kbps when I changed from 375-400QoS. I meant that when I flipped back and forth, 375 dropped down to 123 every time, and 400 jumped up to 232 everytime. Very consistent.

Posts: 482 · Posted: Thu Mar 10, 2005 1:14 pm Post subject:

I thought the whole point of QoS was to tag packets so that they are given priority when the throughput is at its limit.

This information makes it sound like the router is simply segmenting and capping the traffic, even when there's no Voip data in the mix.

If that's the case, that's not QoS at all, right?

Posts: 148 · Vonage Forum Senior Joined: Feb 28, 2005 Posts: 148

Maybe that's why so many people have better luck running the ATA behind a WRT54G with aftermarket firmware that actually does good QoS. I've never been a fan of Linksys. Everything I've owned from them has been absolute junk, and I've RMA'd products so many times it just became cheaper to buy from a competitor than to keep paying the shipping.

Posts: 148 · Vonage Forum Senior Joined: Feb 28, 2005 Posts: 148

WOW, I just changed mine from the default 512 to 1024, and there was a dramatic increase in quality. I am at the 30kpbs setting because of limited bandwidth and quality over my wireless broadband connection. I regularly get voice dropouts, and I'm waiting till later in the month for my ISP to upgrade their connection and switch their upstream provider. However, unless it is some fluke, changing this setting really made a difference in call quality.

Posts: 196 · Posted: Thu Mar 10, 2005 2:03 pm Post subject:

Sandman333 wrote:

Maybe that's why so many people have better luck running the ATA behind a WRT54G with aftermarket firmware that actually does good QoS. I've never been a fan of Linksys. Everything I've owned from them has been absolute junk, and I've RMA'd products so many times it just became cheaper to buy from a competitor than to keep paying the shipping.

I can say the Linksys QoS built in to the firmware isn't great. I upgraded my firmware with the Sveasoft.com firmware and it made a huge difference. If Linksys would explain their algorithms, people might understand it better but I haven't figured out their QoS either.

The 3rd Party firmwares out there can do a much better job on the WRT54GS but there is nothing out there that I know of to run on the RT31P2 or similar devices.

Posts: 244 · Posted: Thu Mar 10, 2005 2:24 pm Post subject:

Thats what I was trying to convey in my last few posts. My setup both in terms of call quality and my overall network health has improved since I put the RT31P2 behind my Sveasoft enhanced WRT54G.

My wife has even asked me what I have done to improve the functionality so Im not the only person to notice it.

Posts: 482 · Posted: Thu Mar 10, 2005 3:08 pm Post subject:

Well I'm not going to go buy another router just for Vonage, but for you guys who can use it, here's where you can download the Sveasoft firmware for free:

http://www.linksysinfo.org/modules.php?name=Downloads&d_op=viewdownload&cid=36