Modem/routers and latency

HarkeN · 3 May 2014 at 21:16

Hi all,

I'm wondering whether or not someone would be able to grace me with some info.

I've recently switched ISP's and was previously using a supplied BEbox as my modem and router (branded TG585v7). I've unlocked it and used it with my current connection (ADSL2+ TTB LLU via Xilo, fastpath and 6dB SNR) and find that I get around 18ms ping to bbc.co.uk with it.

I bought a TP-Link W8961ND as I wanted to ditch the Thomson as well, and find that when I first purchase it, I was getting 12ms - a drastic reduction on the Thomson.

I'm now getting 14ms because for some reason, when I tried setting it up to bridge a connection so I could dial PPPoE from Windows to test it, when I set it back it seems to have increased the latency. I've tried a factory reset and resetting the ROM via telnet (which might be the same thing?) but no dice. Different MTU's/MSS doesn't help either. Anyway.

I've heard that the Infineon-based chipsets in the DrayTek Vigor series will sync best with the TTB DSLAM's, but does anyone have any idea about differences in ping?

Ordinarily I'd have thought that differing pings via different modems would be voodoo nonsense but I've experienced it. I'm aware that a few ms difference is negligible outside of arena FPS games with older netcode, but I'm a perfectionist and I'm curious. Anyone have any stats? There doesn't seem to be a whole lot of info on 'tinterwebs about it.

Caged · 3 May 2014 at 21:22

It's more likely to be that dropping the line to test things has caused the level of interleaving on your line to be increased as it thinks it's a fault. The first router swap could have simply been the new PPPoE session being routed differently.

KIA · 4 May 2014 at 11:23

Did you compare the traceroutes?

HarkeN · 5 May 2014 at 19:39

Yeah, I did. It's not interleaving because it shows as INP=0 and D=1 in the router stats, but the first hop has a higher latency with the different equipment. Unfortunately I don't know if it's taking a different route there because I didn't think to compare.

It was consistent across ISP's as well, as I was migrated to Sky following the BE/O2 takeover and it showed the same behaviour there. Lowest (stable) ping I've had to BBC was 10 on Sky using the TP-Link the week after I bought it (before playing around with it), and the highest was 19 on Xilo using the Thomson. I should mention that the amount the ping increment is identical to every IP, and there was an observable difference in the TBB ping monitor as well, which suggests internal latency of some kind.

I mean, I'm not that bothered, it just seems curious.

HangTime · 8 May 2014 at 21:22

HarkeN said:
Ordinarily I'd have thought that differing pings via different modems would be voodoo nonsense but I've experienced it. I'm aware that a few ms difference is negligible outside of arena FPS games with older netcode, but I'm a perfectionist and I'm curious. Anyone have any stats? There doesn't seem to be a whole lot of info on 'tinterwebs about it.

It definitely can make a difference - I found for example that my Linksys WAG354 gave around 2ms lower latency on average than my Netgear DG834 and whatever Thomson router it was my dad was supplied by Eclipse. I did quite a lot of testing, not just a case of running a couple of pings before anyone suggests this was just natural variation.