Menu
Reply
  • 548
  • 3
  • 31
arronlowley
Up to speed
125 Views
Message 1 of 4
Flag for a moderator

always wondered, why is vdsl better for ping than cable?

i dont understand it at all, cable offers a much better throughput that the standard twisted crap bt and the rest use, but the latency provided by vdsl is generally better, since being with virgin, the speeds have been more than fantastic, but the ping does seem to be all over, i dont mean like into the 100s, but one min it could be 18, then it would be 56, then 18 then 43. ping monitor even shows the spikes. so the question is, why does vdsl offer a vastly more consistent latency. (i used to get 10ms on bt, speed was always dropping though)

 

http://www.thinkbroadband.com/ping/share/6c7f9a9f20c4b3cac9edc401f62167c3.html
0 Kudos
Reply
  • 8.39K
  • 413
  • 987
jb66
Alessandro Volta
110 Views
Message 2 of 4
Flag for a moderator

Re: always wondered, why is vdsl better for ping than cable?

I imagine its because the hubs lock onto 8+ different frequencies to get the high speed. The data gets broken up into 8 slices then when the data gets to the other end the data gets put back together, this must take a few ms to do
0 Kudos
Reply
  • 548
  • 3
  • 31
arronlowley
Up to speed
91 Views
Message 3 of 4
Flag for a moderator

Re: always wondered, why is vdsl better for ping than cable?

would make sense yeah, i wonder if there will be a time where cable is able to achieve faster speeds with less channels.

http://www.thinkbroadband.com/ping/share/6c7f9a9f20c4b3cac9edc401f62167c3.html
0 Kudos
Reply
  • 246
  • 2
  • 23
watercooled
Up to speed
87 Views
Message 4 of 4
Flag for a moderator

Re: always wondered, why is vdsl better for ping than cable?

Part of the reason comes from how DOCSIS (cable) deals with upstream traffic.

At a basic level, it's a collision-based protocol because how how the network is a shared medium a bit like WiFi, cellular, etc - when your modem wants to transmit it risks stepping on (or getting stepped on) by other modems in the service group transmitting at the same time - this is a collision and when it happens, all modems transmitting on that channel have to back off and try again after a short but random time. The random time is a fair way to avoid it happening over and over - the one that picks the shortest time will transmit and the others will 'hear' this and wait for the next gap to try again.

With DOCSIS, there's a bit more to it than this and you have different time slots assigned to e.g. 'requests' and others to actual data packets. This is essentially what you see in the 'upstream burst' section of your superhub. When your modem wants to upload something, it has to request a time slot from the CMTS in the collision-based req slot, then wait for a time slot to start transmitting the real Internet traffic.

As you can imagine, this adds some amount of latency and jitter (variability) to the upstream side of DOCSIS. In general it shouldn't be huge though unless perhaps the service group is getting a bit congested.

In theory, the channel bonding shouldn't add to latency, this sort of thing is done in microseconds, and no different to the OFDM modulation used on VDSL, or how gigabit Ethernet splits data into four streams to send over four twisted pairs.

0 Kudos
Reply