Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Alex Atkin UK]

The only thing I can think is seeing the line rate does give you a guideline for setting QoS on your router.  Other than that, I'm personally looking forward to never needing to see the stats again.

[Robbie]

Sounds like routing is all sorted and a modem wouldn't provide the ability to set static routes over CLI anyway as it's a modem - link layer only.

If you're bereft of line stats and tweaking via G.fast you're really not going to like FTTP 

As you know, the more popular aftermarket 'modems' are really routers in disguise; depending on firmware they usually have their CLI exposed too.

For VDSL I used a Vigor 130 and with it only having a single LAN port the accessing of its web GUI, CLI or line stats offered a small hurdle to overcome.  Using the modem's CLI a set a route along the lines of:

Code: [Select]

ip route add 10.0.1.0 255.255.255.0 10.0.0.1 static back to my router to allow easy access to the modem from the LAN side of the router.  It could be done via NAT masquerade too, but a simple route via CLI is arguably more elegant.  I didn't run the standard BT SIN firmware on the modem either as the alternate loads played better with my ECI cab.

It looks like my optimism for using improved or tweaked firmware for G.fast may have been misplaced.  As to the agony of having FTTP, well that would be a pain I could live with(!) but alas I don't think it will trouble my area for a number of years.

Still very early days for my G.fast experience & hoping the line 'training' and interleaved profile will end soon.

[re0]

What were your estimates and what is your current speed? How far away are you from the cabinet?

G.fast starts off as fast as possible. If it has gotten worse since provision, then perhaps your line can't cope with 3 dB SNRM - the DLM can set it anywhere from 3 to 28 dB, in 1 dB steps (downstream and upstream independently of each other).

By the way, to my knowledge, G.fast doesn't use the interleaving. It uses Impulse Noise Protection (INP). There is nothing to really change.

[Robbie]

What were your estimates and what is your current speed? How far away are you from the cabinet?

@re0

100M to cabinet, 140M by wire (hard to say where that extra 40m is given the simple street layout).  All copper, all ducted, no line issues with VDSL with a sync of 89 / 28 Mbps.

G.fast Estimated: 287 / 44.9 Mbps (241.3 / 31.6 Mbps degraded).

Test equipment on installation:

Line: All greens

Maximum Data Rate: 223.016 / 31.932 Mbps
Actual Data Rate: 159.924 / 30.037 Mbps (ie my 160/30 starting package)
SNR: 10.90 / 3.90 dB
Attenuation: 36.1 / 0.0dB
Capacity: 71.7 / 94.1%
Output Power: 0.00 / 4.10 dBm
Interleave Mode
Interleave Depth: 5 / 4
Bitswap: N/A / N/A
Trellis: On / On
G.INP: Active / Active

Post installation modem sync shown as 144.8 / 28.6 Mbps (via ISP).

BT Availability Checker shows a G.fast Max Observed of 246.25 / 33.31 Mbps (dated 2 days after install)

Current speed test circa 141 / 29 Mbps; latency 15ms (was 8ms on VDSL).

Now at install +10 days, last full modem sync was 8 days ago.

[re0]

I have huge doubts your line is 140m with an attenuation of 36.1 dB. It's probably closer to 250m given those estimates and speeds. I think mine is around 180-190m and my attenuation is somewhere between 29-30 dB.

Which ISP were you on and which are you on now? The differences in latency could be due to different routing used. I am on Zen and the latency can vary up to around 6ms in the worst case - historically, I have had to drop the PPPoE session and connect over and over until I've got the latency/routing I wanted (using ping -t on the host of choice). Have you tried a tracert? Do you have a tracert from your previous ISP/connection that you can compare?

Looks like you should be syncing at the maximum rate for the package, with decent overhead for the downstream (you have spare margin). I know the line data for Zen shows the sync. rate for my line to be about 18 Mbps below the actual sync. rate.

[Robbie]

Between my demarcation point and the cabinet via the ducting route is 131m, when measured on Google maps.  The Openreach laptop had the line length at 140m.  The cabinet can be seen from my front windows as it is almost directly in front of us.

I don't think I have a ping plot from vdsl now, but I will check tomorrow.  The BT availability checker did have a higher estimate for our property, but that was reduced by a good chunk in the weeks leading up to the switch.

There are 9 houses on my road, all served by the same cab.  I'm the first with g.fast so nobody to compare with.  Very few g.fast installations in my area. and the guy doing the instal said he was the only local guy who had done the course.  Indeed, I was his first modem-only install.

Previous ISP was BT, current is iDNET.

[bkehoe]

I have huge doubts your line is 140m with an attenuation of 36.1 dB. It's probably closer to 250m given those estimates and speeds. I think mine is around 180-190m and my attenuation is somewhere between 29-30 dB.

Which ISP were you on and which are you on now? The differences in latency could be due to different routing used. I am on Zen and the latency can vary up to around 6ms in the worst case - historically, I have had to drop the PPPoE session and connect over and over until I've got the latency/routing I wanted (using ping -t on the host of choice). Have you tried a tracert? Do you have a tracert from your previous ISP/connection that you can compare?

Looks like you should be syncing at the maximum rate for the package, with decent overhead for the downstream (you have spare margin). I know the line data for Zen shows the sync. rate for my line to be about 18 Mbps below the actual sync. rate.

I agree, my attenuation is 27dB and I'm about 150m from the PCP with how the cable is routed (and have an attainable of around 400/73 most of the time). If you are sure of the distance that the cable takes then somethings up with your line!

As others have said there isn't interleaving on g.fast so latency should be almost equivalent to FTTP. I get approx 8ms from Manchester area to London on BT.

[Robbie]

I agree, my attenuation is 27dB and I'm about 150m from the PCP with how the cable is routed (and have an attainable of around 400/73 most of the time). If you are sure of the distance that the cable takes then somethings up with your line!

As others have said there isn't interleaving on g.fast so latency should be almost equivalent to FTTP. I get approx 8ms from Manchester area to London on BT.

Thanks guys.  When the Openreach guy showed me the routing of the ducts on his laptop nothing jumped out at me as unusual with the routing but the graphic was not rich with detail - the rough routing is annotated below.  The ducting path depicted is around 131m, direct to cab is just under 120m.  The cabinet itself is under the trees at the top / northern point of the picture:



So as I understand it the attenuation is too high for my apparent line length and given that my speed is well-below expected levels the logic is totally sound.  I’d be surprised to find another 110m in the ducting length somewhere though, even with a healthy error margin - it’s just a cul-de-sac with 9 houses.  I’m home later so shall have a wander around to look for the BT manhole covers.

I re-synced the modem this morning (install +11 days) with 8.5 days since last restart.  No change in reported sync rate from my ISP.

Ping plot from this morning:



G.fast Max Observed:



Install Line Test:



Install G.fast Tests:

[re0]

Well, attenuation cannot be accurately translated into distance, and vice-versa - at least not practically. Though it can be pretty accurate if you know the path, grade of copper/aluminium, etc.

I'd suspect that it's possible that the extra approx. 100m in [electrical] length is due to aluminium and/or thin copper, along with the route not being so direct.

I suppose when the engineer came for install, they renewed your master socket? Are you using the master socket and not a slave socket?

I can't really see your trace, but perhaps you could do some testing and see if dropping and reconnecting PPPoE gives you better/worse latency/routing. And try it a few times.

By the way, I made a mistake. Interleaving is applied, but delay is not - at least not by default, and I didn't see anything in the SIN about interleave delay being used.

[Robbie]

It's all copper, thankfully, and around 13 years old.

Yep, new master socket with the modem directly on the IDC pair, with around 60cm of cat5e to the modem sitting on the floor under a set of drawers - all in an effort to make it wife-friendly.

[Robbie]

The back of the socket is the external wall with the BT wiring entry point from the ducting, so very little extra wiring:

[bkehoe]

Just a thought but sounds like your ISP is showing you the IP Profile of your line rather than the sync rate itself. Your speed test results seem perfectly normal for a 160MBit sync, albeit if you were wanting a faster package at some stage then of course the attenuation is an issue.

The fastest speed tests I get on my 330MBit sync is around 291MBit and thats only by picking a server that has the capacity for example.

[re0]

Absolutely what I was thinking. Speeds are normal for the sync. According to the checker (max observed speeds), the most he could expect anyway would be about 246/33 Mbps if he upgraded.

The mystery of the distance may remain a mystery, but it's safe to say there is nothing that can be done about it unless you wanted to pay for an alternative solution (FTTP, leased line, etc.).

As for the latency, the difference in 8ms vs 15ms is very small on a larger scale. What I can say with confidence is that G.fast is not adding a delay to it - in theory, it can be slightly quicker than VDSL2. The delay is certainly coming from beyond the DSLAM, which is down to how your connection to the internet is routed with your ISP.

[Robbie]

All of the unexpected / additional latency is on hop 2, so noteworthy I think.

ISP provided info: Sync Speed - 144.8 Mbps Down / 28.6 Mbps Up
- Gives me circa 140.9 Mbps on the usual speed testers, so seems about right from a headline 144.8 Mbps sync speed.



I've dropped a note to my ISP.  The package plan was to start on the 160/30 plan and then migrate upwards as you can move up but not down in the first year - an ISP recommendation at order time in order to see the actual line conditions.

After years of VDSL this is all learning for me.

[bkehoe]

Robbie, I'm 99% sure they're quoting you the IP profile rather than sync speed.

I've just checked the BTW speed test and on my 330 sync the IP profile is 303.13 and as I've mentioned before tests max out not much more than 290mbit which all makes sense so what you're seeing on your 160 package implies a full 160 sync.