What if I don't want the NBN... *shock horror*

Well that's shithouse... so they are going to take my decently fast, reliable, cable based internet (which also runs our Foxtel Go/Play), and fuck it up by putting it onto a copper pair service, which (whenever it rains), often results in a line having issues.

Fantastic!!

I know a number of people / businesses on the NBN and not one of them has a positive thing to say about it.

Up NSW Central Coast there are people who had decent ADSL connections disconnected near an exchange, who now get 10 Mbit down / 0.5 Mbit up, due to over subscription/congestion.

In other areas constant dropouts and packet loss.

They should have left the cable network as is and expanded FTTN/FTTH/Fixed Wireless to those homes without internet.

Then again if 5G gets a roll on all of this will have been a pointless exercise anyway. On 4G at home I can often get 65 Mbit down / 40Mbit up.

 

who can say what will happen in the future, but I doubt wireless is our saviour. 5G is great but put it in a contended environment with lots of client devices competing for some very expensive frequency space and the it will have the same congestion issues as 4G does now.

people need to differentiate between causes of issues. FTTN congestion is not an NBN issue. it's 100% an ISP issue.

 

I don't know - we don't pay full price because we host a bunch of their infrastructure. Installation costs are minimal and there's no ongoing contract.

It's cost us about $8000 since June 2014. There is no alternative at our location - no fibre coverage, from anyone. ADSL speeds were around 10/1, and there's no cable or realistic access to nearby fibre.

 

(since you probably have access to this info) does that mean that none of the nodes on the NBN network have congestion on their 2gbps uplink?

 

what 2Gbps link is this you speak of?

that aside, zero of them. they are watched very carefully.

 

On multiple ISPs same issue? Only common issue is the node!

 

Maybe there is a lot of misinformation floating around, but I was under the impression that there are (currently) 2x 1Gbps links to each node, expandable up to 4 links, but going to 10Gbps would require an additional module in the ISAMs. Of course, you know how things are actually implemented, so feel free to set the record straight

Kenneth Tsangs blog: http://blog.jxeeno.com/poor-nbn-fttnb-design-may-lead-to-decades-of-congestion/

 

given I have direct visibility of the issue, I say it is not with total confidence.

jxeeno has a history of reading factoids and interpreting them incorrectly. I don't blame him entirely for that, he's an amateur and makes amateur mistakes and assumptions.

there are two fibre pairs initially provisioned to each node, each of which can carry industry standard 1Gbps or 10Gbps ethernet links. the "module" change is simply which optical SFP is selected, which is done for every node right now on the basis of required reach.

the node itself can terminate up to 4 links expandable to 10 links with a simple applique board. that means 100Gbps available right now for the asking. if more fibres are required then it's child's play to drag an additional cable in, a single 12F ribbon is less than the size of a pencil and there is considerable network capacity upstream built in to terminate it.

in practical use, there are zero nodes in the network that actually use all of their initial 1Gbps link, or anything near it. most are subrate limited to well below that because real world peak bandwidth demand is about 1/10th of what people think it is.

to illustrate, most greenfields FTTP suburbs are served by a cabinetised OLT (much the same electronics, just fitted with optical instead of electrical line cards), and they all have an initial, single 1Gbps link too. those OLTs can accomodate well over 5 times the connections of a single FTTN node, with the additional considerations that:

• every premises served in a greenfields estate has to use a fibre link for a service, there's no remaining on copper for the moment because no copper ever existed
• every premises can get a 100Mbps link on demand with zero consideration for DSL performance over long copper runs

and yet most of them still don't need more than their 1Gbps link, in many cases subrate limited, at considerably more than 5 times the aggregate peak bandwidth demand of a comparable FTTN node.

on top of that, there are active monitoring systems and an entire department which look over uplink bandwidth, and plan augmentations around 3 months before projected future usage might reach saturation levels. that might mean anything from ordering an upgrade of an additional link, to ordering a parallel link for capacity reasons, to migration to a 10Gbps channel.

this isn't Telstra all over again, where network congestion was allowed to occur quite deliberately and with full visibility, by management decision. the NBN standards for network uplink performance and upgrades are published and available not only for all of the NBN access seekers, but for the public to see too.

I can think of perhaps three, perhaps four instances in the last five years when uplink utilisation has meant potential service impact, and then only because managed backhaul providers dropped the ball and couldn't deliver on a planned augmentation in contractual timeframes. it's a subject that is watched very, very carefully.

so to summarise - if you're on a terrestrial NBN connection, and if you experience peak hour congestion, it's inadequate CVC at fault and that's the end of it. period.

if you want a litmus test, log a fault with your ISP. they will inquire with NBN. if it's NBN's fault they will relay the blame on immediately, I am sure. see what answer you get. I've done a number of drilldowns on the subject where the content of the investigation has ended up in a submission to the cabinet estimates committee. so far I'm batting 100% and I have no expectation that will be changing.

 

Fair enough, sounds like a good deal for you then. But for anyone who wasn't acting as a host and getting those discounts, if the TPG 400 service was available, unlimited 400Mbps symmetrical at $399/mth (plus a hefty $2k+ install fee) is a better deal - and it's a fixed line service.

 

There are always options, you just can't justify them commercially. Any of the top 4 (and some below) can and do build out their assets on demand providing a client is willing to pay the $$. Don't equate "I can't afford it" with "There isn't any coverage". Some businesses would give their left nut to pay $8000 a month for 100/100Mbps.

 

Or maybe more than 1 ISP is not buying enough CVC capacity. Since it costs so damn much this is entirely likely.

 

CVC pricing is based on what it costs to run the company, at the government mandated rate of return. there is no making it go away, the cost would just be transferred elsewhere and the net result is zero.

Australians undervalue what a decent broadband connection costs to provide, and making irrelevant comparisons to old ADSL connections or services in another country with a totally different set of delivery costs means nothing. it just confuses the debate by introducing attractive but irrelevant numbers that people then cling to because they'd like them to be true. the same applies to comparing the cheapest price that can be found in the market to a decent connection, and expecting quality at rock bottom pricing, all while complaining about a lack of "competition", which is in reality bitching about the price.

while we continue to balk at the idea of actually paying for what a quality connection costs to provide, ISPs will continue to cut costs, and CVC capacity is the easiest variable they can control.

while we continue to focus on price, all we do is drive quality out of the market. this article puts it well. https://en.wikipedia.org/wiki/The_Market_for_Lemons

if anyone needs a real world example, look at what's happened to the once-premium Internode.

 

Thanks for that great explanation. Looks like a lot of what I've previously read has been incorrect (e.g. the SFPs were limited to 1Gbps without additional cards being installed in the node to support 10Gbps).

I should probably appreciate the diversity between customers more (working for a DNSP). It's good to know that everything within the NBN network is actively monitored.

On a separate note, a node down the road from me was involved in a car accident a couple of weeks back: http://imgur.com/U72x8fr
I was surprised to see the NBN crew out the next day, installing and commisioning a new node. Took them a few days to get it back up and running (I guess a new link to the pillar adjacent would have been required, as well as jumpering them all). Props to them for a quick response.

 

there were three that week, if that is the site I am thinking of. >_< unfortunately death by vehicle is a part of life for roadside kit, and the team are getting used to it.

the alarm monitoring team know in about 2 minutes that something is wrong and they roll a truck, but fixing the issue is a combination of hard work by the field teams, logistics to get the required parts kit to site, and reconfiguration of the gear.

 

That's FTTNope.

 

ITT: aokman doesn't support the most important nation building project of our generation because it doesn't help him right this second...

Oh shit, I bit!

Not sure why you think you can't use your Fritzbox.

And not much of your story checks out. Maybe you've just been sold a lot of tripe by a TPG sales rep? When your ADSL connection goes, ISPs null the contract as they can't supply their end of it anymore.

 

Excuse my dumb question:

Are CVC's like ports on a DSLAM? I.E Limited in number per exchange?

Which ISPs have the most?

 

a CVC is a logical "container" of bandwidth the RSP buys inside the NBN network. it's a combination a VLAN address unique to the RSP, desired peak bandwidth, and what QoS grades they want to offer. some QoS grades are 1:1 protected, others can be overbooked (best-effort). most residential grade data connections will be the latter, so that's what I am describing.

the CVC is purchased per-POI, and carries every individual service connection the RSP has at that POI. they can buy multiple CVCs if they want or need.

I found a reasonable diagram explaining this here, the CVC exists between the EFS and EAS boxes. it's a bit small but zooms well enough to still read.

https://image.slidesharecdn.com/oli...02/95/olivier-duroyon-8-638.jpg?cb=1362507809

on the left-hand side, where the access DSLAMs and OLTs are, there could be several hundred of these devices per POI, each with anything up to ~4,000 users for an OLT and around ~350 per DSLAM. obviously the RSP doesn't serve all of these though, so it's up to them to figure out the number of services they have sold per POI, the mix of speed tiers, and the average peak bandwidth demand of them all (which is generally a small fraction of what people think it is). obviously some sorts of internet usage (e.g. multimedia streaming and P2P) can generate sustained, high bandwidth usage compared to the more bursty usage of other use, so the RSP needs to take this into account too.

if the RSP allows peak bandwidth demand for the POI to exceed their CVC bandwidth capacity, every user on every sort of access technology for that POI will perceive it as congestion. this poses a challenge for the RSP. like freeways that are only really used to maximum capacity for a couple of hours twice a day, or shipping centre carparks that are smashed for the fortnight before christmas, if the RSP provisions enough CVC to result in zero congestion at peak usage, they are paying for CVC they don't need for most of the day. that cost flows directly through to your monthly cost as a user. if the RSP goes down the route of allowing some allowable peak hour congestion to occur as a cost control exercise, people bitch about their Netflix buffering while they eat dinner, forgetting that they bought the cheapest plan they could find.

if an RSP is going to compete on the basis of price, they have to control their costs somehow. CVC is where it happens.

IMO everyone involved needs to take a deep breath and accept the fact that an NBN connection is going to cost more. consumers will pay higher prices. RSPs need to price their products at a rate that allows them to work properly. and we all need to stop trying to make prices match those of legacy ADSL - we all complain about the performance and want something better, making that happen costs money.

 

Thanks for the awesome reply and in depth explanation.. appreciate it.

So it's a bandwidth pipe (much like say Exetel offering fibre services, but on a shared 1Gbit link from AAPT/TPG)?

If that's the case, it explains why "business NBN" (according to most ISPS) still isn't a thing per se.

On the user side of things: If users never experienced this with ADSL or their cable (I know people that are on both technologies converted to NBN), then does this mean the ISPS are cheaping out on the smallest CVC possible and hoping it will "suffice", because they no longer have minimum bandwith requirements to contend with?

I'd have thought someone like Telstra would have more than enough CVC for users in an area.

 

Many RSP's limit their CVC congestion exposure by implementing download limits or peak/off peak qoutas, this allows them to reduce or move the heavy users away from peak periods (3pm-11pm) this allows them to keep resonable speeds for all users.

Regardless the cost of CVC is too high, and rather having a AVC/CVC cost split they should be combined. At least NBN has acknowledged that the cost is too high and is working on getting it lower.