Kitz Forum
Broadband Related => Broadband Technology => Topic started by: ktz392837 on December 14, 2018, 03:48:05 PM
-
I appreciate that there are quite a few variables but if you are trying to estimate if you can benefit from Gfast is there any correlation between the speed of a FTTC connection and the speed of a Gfast connection on the same line?
For example, if you connect at FTTC 67/19 with an attenuation of 17db (ECI cab) what would that equate to if upgraded to Gfast?
If anyone has FTTC stats before they upgraded to Gfast and know their Gfast sync speed could you post them?
Thanks
-
Every post I recall the person already received the full 80Mb.
Why not find an area with live G.Fast and work outwards from the cabinet and look for those with similar FTTC estimates to you.
My initials thoughts are you are probably out of reach of the current generation of G.Fast equipment.
The upstream in particular tends to drop below FTTC estimates at longer distances.
-
Too many variables. Look how much vdsl2 can vary when you have 2 lines with the same attenuation. The higher the frequencies in use, then the more variable any estimates are.
My daughter's line has an insertion loss of 8dB (true attenuation). But line attenuation on g.fast is something like 40dB. They previously got the full FTC 80Mbps and max sync with g.fast is about 240Mbps. Its quite variable though and my daughter isn't too impressed at the line can often drop out. Zen are supposed to be sending them a new router but :shrug2: because they can't seem to decide if its full loss of sync or loss of PPP. Lack of stats from the modem doesn't help, but there are some line stats in the relevant thread which I asked them to get from the Openreach Engineer for me.
-
I would have to agree with both j0hn and kitz. The former having said you are likely out of reach is, in my opinion, almost certainly correct since your line length is probably exceeding 350-400 meters (at which point G.fast is simply unable to offer any benefits and upstream as j0hn said would be worse).
Its quite variable though and my daughter isn't too impressed at the line can often drop out. Zen are supposed to be sending them a new router but :shrug2: because they can't seem to decide if its full loss of sync or loss of PPP. Lack of stats from the modem doesn't help, but there are some line stats in the relevant thread which I asked them to get from the Openreach Engineer for me.
I don't have G.fast but from the limited feedback/stats I have seen it could be said that drops seem to be a lot more common with G.fast than with VDSL (somewhat expected utilising a much larger frequency band with paper thin noise margins). Regardless of whether it resyncs in seconds, it is probably frustrating given that VDSL, for most people, can stay synced for days, weeks or, like the case here, even months.
Hopefully G.fast won't turn into G.farce, or perhaps it already was from the start. ;)
-
Thanks for replies.
As suspected I am out of reach of gfast especially as I pay for 80/20 for the faster upload which will probably be downgraded on gfast.
The only benefit might be crosstalkers moving to gfast and making my line return to 80/20 especially as ginp/3db is not happening.
That's assuming gfast is ever deployed for me. Guess I am stuck with the eci cab. Just hoping I don't end up at my bottom estimates again which are so bad I would need to downgrade to a lower package.
-
Nobody will be selling it to you if all the estimates are under 100 Mbps anyway. It won't even show on the DSL checker. There's always the slim chance that you may be on the edge of the coverage but without much of a doubt it will be at a cost to your upstream. Some examples can be seen here (http://www.optanet.com/ultrafast-broadband-checker/) and here (https://aastatus.net/2529), though it further proves that the upstream can be extremely enemic at further distances.
I did have a thought, and that is G.fast could actually be faster than VDSL for a very small number of people who are outside the range to receive 100 Mbps (approx. 350-450 meters, but still receiving ~70 Mbps on VDSL) meeting the following critera:
- ECI cabinet - absence of XdB and G.INP (and vectoring in some BDUK areas)
- Large number of subscribers (a lot of crosstalk)
Though the stability and consistency of performance can be questioned.
But fingers crossed that those eligible will upgrade so you can get a slight speed bump. ;D
-
When lower end estimates get down below 100Mb then instead of available the line shows as "Amber".
Houses a considerable distance from the cabinet don't show G.Fast at all.
LS28 5JA is a good example of G.Fast with a wide range of high/low estimates.
It's part of P55 on Pudsey exchange (MYPUD).
I don't see a single instance of G.Fast available where the high estimate for VDSL2 isn't 80Mb.
-
I'm failing to see the point of G.Fast if it's only going to be for those who can get 80Mbps anyway.
A sizeable proportion of the cabinets with G.Fast will, in a few years, have many of the users outside of the 80Mbps zone moving to FTTP. As people realise what the full benefits of a genuine fibre connection (as opposed to hybrid fibre) are, people who can't get FTTP because they have G.fast available are not going to be impressed. So why divert energy in to what can only be seen as a temporary stop-gap? Crack on with FTTP I say.
-
I can understand why you have that opinion. It is not necessarily an unpopular opinion to have on the forum. VDSL 35b would have certainly been a better upgrade path compared to the idea of mounting G.fast pods to cabinets as opposed to distribution points. The current amendment and possibly even later amendments are not going to be as effective as the standard could allow because of the issue of co-existence and effective network arrangement to allow for pods closer to the customer (and at such a point, why not just FTTP?).
Bearing in mind that it's probable that some of the shortcomings may have influenced their decision to reduce planned future G.fast footprint to some degree in favour of more FTTP. In some denser areas, it may be possible to cover in excess of 70-80% of premises with 100+ Mbps with G.fast eligibility but the reality is that areas that had plans scrapped probably were not very good candidates as they may have been sparse or had cabinet coverage where there copper network rearrangements would have been necessary to give reasonable coverage; we all know too well that sometimes addresses or even whole streets are connected to cabinets several hundreds of meters away when they might even be right next to one that serves the surrounding area.
I don't know for sure so it's all just a bit of speculation. Pinch of salt advised.
-
I did have a thought, and that is G.fast could actually be faster than VDSL for a very small number of people who are outside the range to receive 100 Mbps (approx. 350-450 meters, but still receiving ~70 Mbps on VDSL) meeting the following critera:
- ECI cabinet - absence of XdB and G.INP (and vectoring in some BDUK areas)
- Large number of subscribers (a lot of crosstalk)
My PCP is also 30m closer than my FTTC cab (perhaps longer depending on cable routing). I also have a massive range on my estimates (25Mb) so I would actually meet the gfast only for current 80Mb fttc criteria. Going off what everyone is saying though it is probably not going to be worth the risk even if it does become available.
-
ECI cabinet - absence of XdB and G.INP (and vectoring in some BDUK areas)
Missed this part.
BDUK exclusively used Huawei cabinets.
-
Likely off original topic but within the current topic trend:
Surely more extensive roll out of vectoring would have produced improved VDSL2, FTTC performance to a much larger percentage of users than G.Fast in it's present format is ever likely to do, obviously accepting the 80 Mbps upper speed limit.
I can fully appreciate that vectoring is not perceived to be a very good business prospect to BT, considerable financial investment in hardware and man power, FTTC cabinet compatability issues etc. with no quick way to recoup some of the outlay with increased tariff as per G.Fast, realistically, never likely to happen now.
-
When Openreach first announced g.fast, I think a lot of us foresaw benefits of it, because in 2017 they said they were trialling g.fast using FTTdp. IMHO this would have been progression to bringing full fibre one step closer.
The fact they only seem to be deploying from the cab means that only a small portion of users will be able to benefit from g.fast (ie those already getting 80Mbps) :(
-
Missed this part.
BDUK exclusively used Huawei cabinets.
In my haste ... ;) Thanks for adding it.
Surely more extensive roll out of vectoring would have produced improved VDSL2, FTTC performance to a much larger percentage of users than G.Fast in it's present format is ever likely to do, obviously accepting the 80 Mbps upper speed limit.
Vectored 17a would have almost doubled data rates over standard 17a, providing ultrafast (100+ Mbps) to around 500m. I won't go into detail, but I would love to rant about how 35b would have been so much better than G.fast to cover with ultrafast speeds (and able to co-exist with 17a) and give even 300+ Mbps to approx. the same distance. Deutsche Telekom use 35b.
When Openreach first announced g.fast, I think a lot of us foresaw benefits of it, because in 2017 they said they were trialling g.fast using FTTdp. IMHO this would have been progression to bringing full fibre one step closer.
The fact they only seem to be deploying from the cab means that only a small portion of users will be able to benefit from g.fast (ie those already getting 80Mbps) :(
Quick 'n' dirty deployment. Bolt 'em up and watch 'em go. I can't remember if we discussed it on the forum before regarding reverse power, but perhaps it may have been part of the reason.
Anyway, I feel bad for brining this away from the intended topic. :-[
-
Are you sure?
Profile 17a is limited to 100Mb, despite attainables offering way more.
Vectored 17a would have given "up to" 100Mb.
-
Are you sure profile 17a is "limited to 100Mb"?
The G.993.2 document specifies minimum mandatory requirements. It specifies a "minimum bidirectional net data rate capability" of 100Mb for profile 17a. That's not a maximum limit, it's a requirement for a minimum capability of 100Mb for the sum of upstream and downstream bandwidth.
-
I was looking at this (https://www.researchgate.net/figure/Downstream-bit-rates-for-vectored-and-regular-VDSL2-Profile-17a-bandplan-EU32_fig1_224180554) which shows data rates in the case of Vectored 17a (bandplan EU32).
Connections in Germany using VDSL 17a G.Vector appear to be able exceed 100 Mbps downstream, seemingly going up to 109,344 Kbps. Combined with the upstream being upto and in excess of 40 Mbps, you end up with ~150 Mbps bi-directional.
Perhaps the general limitation is chipsets/DSLAMs supporting upto 100/50 (DS/US) Mbps on profile 17a? I don't know.
-
I was under the impression the maximum combined was 140Mb for some reason.
I wasn't aware Germany had 110/42. Very little online about EU-32 bandplan.
I can't for the life of me find any numbers for what bands they use and what tones.
That's still a very long way from
Vectored 17a would have almost doubled data rates over standard 17a
Most telcos who sell Vectored VDSL2 17a have 100Mb downstream.
-
I wasn't aware Germany had 110/42.
Well, the thing is that it's not actually advertised as 110 Mbps. Both 1&1 and Telekom (Deutsche Telekom) advertise their speeds as up to 100/40 Mbps, but it appears that the DSLAM and modems when using 17a G.Vector seem to be able to acheive sync rate of up to 109344/41999 Kbps (DSLAM says 42000 is max. upstream rate, but I have not seen any modems at that exact figure). Perhaps the 100/40 Mbps refers to maximal throughput, but I somehow doubt it.
A guy who calls himself wunsel on the blogdoch blog did a "shootout" with various Fritz!Box VDSL modems which can be seen here (https://v2.blogdoch.net/2017/05/31/wusel-095737/). It shows both the 7412 and 7490 hitting about as a high as they can go.
That's still a very long way from
Vectored 17a would have almost doubled data rates over standard 17a
I totally agree that would be true, and this was an oversight of mine when I initially wrote that out.
Most telcos who sell Vectored VDSL2 17a have 100Mb downstream.
Perhaps for practicality it would make sense just to say 100 Mbps downstream since the amount of devices capable of syncing at such speeds may be limited.
Either way, I think this should have all been it's own topic.
-
Related to the topic, my lines are 600m in distance from the cabinet, and G.Fast is not available to me at all. I get around 70Mbit on each line.
G.Fast, whilst very nice for those who can get it - only those who have 80Mb can take advantage. Most users who have 80Mbit will likely argue that they won't need anything faster than that.
G.Fast would have made slightly more sense if BT actually pushed on and deployed it in the manner they originally intended it - that is to the distribution points. In my case that would have put me at the 330Mbit - I can see the DP from my house, it's 3 doors away.
-
I don't think the intention was ever to put them on every DP.
It's a big assumption they would have used your DP.
It may have been 2 or 3 DP's along but still would be better than a cabinet deployment.
I'll be gutted if my PCP gets a G.Farce pod.
None on my exchange at all fortunately.
My PCP is the most populated on my exchange, serving over 850 properties.
It would almost certainly get a pod if they run it out here.
-
It may have been 2 or 3 DP's along but still would be better than a cabinet deployment.
Wouldn't that have necessitated network rearrangements to the existing copper networking that Openreach probably would like to avoid?
It may be an obvious answer to you or someone, but it is a question of how signal injection is going to work for those on other DPs without rearrangement.
-
You've lost me.
I don't see how that wouldn't work.
I don't think the intention was EVER to site a G.Farce DSLAM/Node on every DP.
They are installing 1 line card (24 ports) for an entire PCP.
Even a small 4 port pod for each DP would be excessive capacity for initial expected take up.
My street for example, a cul-de-sac.
There's about 6 Dps on my side of the street.
Let's number them 1-6 with 1 at the start of the street and 6 at the far end.
They could place a pod next to DP 1 that serves DP 1 to 3.
Remember it's fibre to the pod so any copper prior to the pod becomes E-Side and the xDSL signal is filtered from the E-Side.
The next pod is placed on DP 4 to serve DP's 4 to 6.
Again they use filters so the G.Fast signal only goes down the D-Side.
My street is only about 150-200m long though so even a single Node on DP1 could do the whole side of the street.
This becomes slightly more complicated in areas with multiple branches feeding of on multiple directions.
If they can filter the VDSL2 signal at the cabinet so it doesn't go down the E-Side then I'm at a loss to see how they can't do the same deeper in the network.
edit: I was thinking abbot this further and I wasn't so sure it would work. I'm confident it would though.
I wouldn't mind hearing other people's theories on this.
I'm also curious if anyone else had thoughts on how OpenReach were originally going to roll out G.Farce.
Did anyone else pick up that there would be a node on every DP or that they would be more selective than that?
My initial thoughts are there are over 4 million Distribution Points and that just wouldn't be feasible.
If there are 5 DP's within 100m of each other then why not strategically place 1 node to cover them all?
-
I agree with kitz, vectored 35b rollout would have been better, sadly tho I think openreach chose g.fast because the max possible speeds are higher, and they went with what has the best impact on marketing.
But vectored 35b to all would have helped many more lines.
-
I agree with kitz, vectored 35b rollout would have been better
That was re0.
-
35b (Super-Vectoring) up to 600m, vectored 17a >600m. 35b and 17a utilise same tone spacing and so the vectoring can be done between them both.
I can't remember whether it was mentioned before on the forum, but the MA5603T does support up to 384 SuperVectored ports. Though it's subject to line cards and controller units which I can't remember details for. :-X
Too bad for ECI deployments since ECI M41 is apparently a nightmare for vectoring. V41 would be better, though no sign of 35b support.
-
then i agree with re0 j0hn
-
As far as I understood it, G.Fast was going to be installed at pretty much every DP. My DP serves at least 64 properties, so I don't think it to be some wild assumption they'd skip over mine.
Seeing as I am only 600m from the cabinet there probably aren't that many DP's between my house and it - our street is the first cul de sac street away from the cabinet.
-
60 properties on a single DP is NOT the norm.
I'm on a brand new build and every DP has 10-12 properties.
Looking over my back fence at the existing telegraph poles there's 10-15 per DP.
edit to add again: 4 million DP's on the OpenReach network, divided by homes. The average is considerably smaller than 60.
I didn't even know there were DP's with 60 properties, unless it's a huge block of flats/apartments.
-
I was going off the assumption that eventually they would install some more pods further into the network, anything other than that would be the work of a moron.
e.g POD at Cab then 400m or so later if it makes sense add a cab at a joint.
-
60 properties on a single DP is NOT the norm.
I'm on a brand new build and every DP has 10-12 properties.
Looking over my back fence at the existing telegraph poles there's 10-15 per DP.
edit to add again: 4 million DP's on the OpenReach network, divided by homes. The average is considerably smaller than 60.
I didn't even know there were DP's with 60 properties, unless it's a huge block of flats/apartments.
OK fair enough, there are many DP's in my street.
-
I agree with kitz, vectored 35b rollout would have been better, sadly tho I think openreach chose g.fast because the max possible speeds are higher, and they went with what has the best impact on marketing.
But vectored 35b to all would have helped many more lines.
Surely, I cannot understand of why isn't Openreach trial G.Fast / VDSL2 Profile 35b with vectored at the same time to see who would benefits better in market sales.
-
Surely, I cannot understand of why isn't Openreach trial G.Fast / VDSL2 Profile 35b with vectored at the same time to see who would benefits better in market sales.
Quote from Reply 11 this thread:
I can fully appreciate that vectoring is not perceived to be a very good business prospect to BT, considerable financial investment in hardware and man power, FTTC cabinet compatability issues etc. with no quick way to recoup some of the outlay with increased tariff as per G.Fast, realistically, never likely to happen now.
-
It's not quite so simple Max.
35b increases speeds by opening up more bandwidth at higher frequencies, but it also relies heavily on vectoring.
It also works best on newly constructed sites ie those without any other DSL services. To quote Huawei
In a reconstructed site, the original services and users interfere with new SuperVector services and users and the interference cannot be eliminated.
Ideally all the users and services should be brand new. Although it is possible to reconstruct a site for 35b all users lines must use SuperVectoring. What are you going to do with the LLU adsl2+ lines. The likes of Sky & TT would have a hissy fit and OFCOM would likely pander to them.
You are also forgetting that whilst the Huawei MA5616 are classed as being upgradable for [normal] vectoring, you practically have to rebuild the DSLAM with new & replacements parts. As I've previously said in another thread (https://forum.kitz.co.uk/index.php/topic,22764.msg387349.html#msg387349), it would be easier (& cheaper) to upgrade the ECI M41s to V41s than it is to rebuild a Huawei MA5616 in the field.
-
IMHO a large part of the reason the existing g.fast isn't working out as well as it could do is due to restrictions from legacy DSL systems.
When you have various technologies working along side each other, using the same lower frequencies, it becomes inevitable that the older technologies gets priority on the lower tones... which are the ones where most of the bitload is usually available. PCB and PSD masks are needed to ensure that g.fast and vdsl doesn't drown out legacy dsl.. which in turn means that they don't get full benefit of the lines full capability.
-
Think in the case of G.fast it's not power masking that's the problem, more that it can only use 20-106 MHz with the current set up. Spot on that this is for compatibility with legacy services though.
ADSL does need to DIAF where hybrid services are available for sure.