Which WAN packet overhead for Telus VDSL2? (Canada)

[BikeHelmet]

Hi all,

I'm setting up a Merlin router for somewhere and I'm having trouble figuring out which setting is the correct one for FlexQOS v125 WAN Packet overhead. It's Telus VDSL2 running through a T3200M which will be bridged.

The two VDSL settings provide values of 22 and 30. I may also use VLANs to segregate the network and keep VOIP stuff totally separate. Should I use that option instead? 42?

If anyone has an idea, I'm all ears.

Cheers,

 

[joe scian]

makes no difference at all. you can enter 0 and it will work just the same

 

[BikeHelmet]

makes no difference at all. you can enter 0 and it will work just the same

Really? It doesn't make any difference? Is there something different about that VDSL2 setup, or does it just "not matter much" anymore?

I typically have quite a few downloads going, lots of small connections, so I know a little bit can add up fast.

 

[BikeHelmet]

Turns out VDSL2 is basically perfect ethernet encapsulation tech. No PPPoE, no frames, no "divisible by" weirdness or anything like that. The modems sync up at a rate higher than your provided speed. You have 30mbit? Maybe your sync rate is 33mbit... 34... 35.... whatever speed is needed and negotiated by the VDSL2 hardware on both ends, to get a solid connection without frequent retransmits.

Your router sends the modem 1500 bytes, it pads it to 1652...1676... 1708 (ECC and transport overhead)... whatever number, it gets to the other side, gets stripped, then 1500 bytes get sent on down the line. There's no splitting up of packets because the VDSL2 Modem and DSLAM effectively don't have an MTU - they transmit the length needed with all the overhead to get 1500 bytes to the other side. So any number than 0 for overhead and 1500 for WAN MTU is wasteful - unless you plan on transmitting VLANs down the line... which apparently it natively supports... then you might possibly need to account for the VLAN header. (Though I'm not 100% clear on that.)

While going down the rabbit hole I learned a lot about PPPoE and also not to measure different things and assume they are the same - like encapsulation overhead vs final destination header overhead.

Ping Test to determine Optimal MTU Size on Router

Use the command prompt to send large packets of data outbound, gradually trying smaller packets until you get a positive reply to determine the optimal MTU size.

kb.netgear.com

How can I optimize PPPoE connections? | SonicWall

Configuring the interface with an incorrect MTU size can cause some websites to not load and Internet speed to decrease. This article describes what is the optimal MTU size for PPPoE connections and its reason.

www.sonicwall.com

Wan Packet Overhead For Qos

How do i calcualte the wan packet ovherhead for a vdsl2 connection llc based or vc based bridge mode, but its ipoe (dhcp) encapsulation not pppoe, just need it for qos.

www.snbforums.com

[Internet] Per packet overhead on Bell's VDSL? ATM based? - Bell Canada | DSLReports Forums

Forum discussion: Does anyone happen to know what the per packet overhead is on Bell's VDSL deployment? It appears from some of the other threads that this service is not ATM based but I'm not certain this doesn't vary by the DSLAM. Also, does Bell still use ATM backh

www.dslreports.com

SQM Link Layer Adaptation question (VDSL2)

Thanks everyone for the input. Much appreciated. I am running the tests below with ethernet hooked up directly to my laptop and all other devices are disconnected from the router. At this point I can't get A+ on all three things (quality,bufferbloat and speed). Here is a few things I tried...

forum.openwrt.org

Turns out that I was looking for a problem that wasn't present!

Cheers,

Oh - I also discovered that if you set your WAN Packet overhead wrong, it shows up as BufferBloat on the DSLReports SpeedTest. It reliably picks up on it (for my connection) and shows the upstream stretching into the yellow danger zone, rather than having no such danger zone visible. (Which is every SpeedTest with it set at 0) That suggests that higher numbers are wasting speed, and I'd have to drop my QOS upstream limit if I wanted BufferBloat back in check with the wrong setting.