As you probably know, I have written quite a bit on this topic in the past trying to go beyond the religious war that has errupted over the two topologies and has increased more recently with the NBN annoucement.
It is interesting to read what other analysts have written on the topic. I should point out that I have never come across Keymile but research has indicated they are a Point to Point vendor - so clearly they are going to have a bias.
The paper is short enough that it's a quick read. From a methodology standpoint, it lacks sources, which in this instance and considering how contentious the issue is in the right circles (or the wrong circles, depending on your appetite for tech talk) is probably a failing.
As usual the paper brings up the same old issues of:
- Bandwidth
- Expansion
- Security
- Flexibility
All of which have been proven to be myths with a few simple and more detailed explainations. For an alternative comparision and myth busting you can look at one of my own presentations on PtP v PON.
10 comments:
- At 07 September, 2009 02:04 Rudolf said...
-
I read your presentation and well, what can we say. It is light on sources and high on hyperbole. To paraphrase:
PON is better. PON is better. Really trust me PON is really better.
Look, the price difference between the two topologies is about 10%. (source: WIK) Going PtP doesn't mean you can't use passive optics in the active layer. One operator can choose to go GPON, another can choose to go EPON, or 100mbit/s or 1gbit/s ethernet. However you can't do it the other way round. What you forget is that the choice for a topology is a choice for the coming 30-60 years. A 10 percent cost increase in the passive layer amortized over 30 years is generally quite worth it.
The nice thing of going PtP is also that there doesn't have to be a choice for LLU or bitstream access. Both are possible.
For regulators the question shouldn't be what technology in the active layer do I choose. For regulators the question is: Do my choices as a regulator influence the technologies that can be chosen in the future so much that certain choices become impossible. Regulators shouldn't get involved in this holy war. - At 07 September, 2009 15:29 Stephen Davies said...
-
Hi Rudolf,
Clearly you have never deployed the technology - PtP or PON.
The cost difference between PON and PtP is significantly more than 10% - in fact currently its about 3-4 times more expensive to deploy a large scale PtP network than a PON network when you consider the equipment, the outside plant and the exchanges required to deliver the services.
As for the WIK report it has been roundly debunked at the last European FTTH conference and is way off the mark in many of its costings.
As for light on sources, there are actually a number of references in the presentation - and if you read some of my white papers you can see the references - but the best references are my own.
Having spent 9 years designing and deploying PtP and GPON and EPON technologies there are not too many people around who have the experience in technology, outside plant and exchanges. - At 07 September, 2009 15:31 Stephen Davies said...
-
There is always one question I like to ask the PtP vendors and proponents.
If you are a CLEC, how do you fit 20,000 fibres into a single exchange. Because this is the point of FTTP, you need few exchanges as the reach is far greater on fibre. With less exchanges you service many more homes.
So how do you fit 20,000 fibres into an exchange? - At 07 September, 2009 20:18 Rudolf van der Berg said...
-
So you are saying that Reggefiber in The Netherlands who are building an FTTH-network for under €1000 per household could have done the same for 250 euro?
And how do you arrive at that conclusion? - At 07 September, 2009 21:24 Ger Vloothuis said...
-
Q: So how do you fit 20,000 fibres into an exchange?
A: 20 racks or cabinets with 1000 terminations each. Not much different of terminating 20,000 wire pairs really....
For an exchange serving 20,000 customers that sort of rack space or cable termination space is not excessive at all. And it has the very big advantage that you can actually TDR the line from the exchange right to the subscriber endpoint if you need to locate cable faults. You can also upgrade services to GbE or 10GbE if a customer wants top pay the additional fees, without needing to modify cable plant. - At 08 September, 2009 10:34 Stephen Davies said...
-
Ger,
Firstly if you have every tried to terminate 1000 fibres into a single rack and then still be able to manage those fibres you will soon find its not possible.
Yes you can get rack which will terminate 1000 fibres, but the patch cord management becomes totally impractical.
So great in theory but in practice not possible.
Further more, you have not considered the duct space required to bring in those 20,000 fibres. You will need some 8 (or more) 100mm ducts for this number of fibres. - At 08 September, 2009 10:52 Stephen Davies said...
-
Hi Rudolf,
It is my understanding that Reggefiber in The Netherlands working in conjunction with KPN are using the existing ducts in the ground.
Secondly I doubt if the E$1000 that you quote includes the PtP OLT, ONT, exchange and all fibre infrastructure.
The cost of deploying a FTTP network is not just the fibre or just the equipment, it is the complete end to end solution including cabinets, exchanges, power systems, fibre, joints, cable management and all those ancilary systems which are required to support and make the FTTP solution work.
You also then have to consider the Total Cost of Ownership (TCO) which includes the maintenance and repair of the network ongoing. A high fibre count system such as PtP has a much higher cost of maintenance and repair.
Furthermore it is my understanding that Reggie is also only delivering fibre infrastructure and not services (I could be wrong here) and as such shifts the paradim of costs involved. There are considerably more costs involved to light up that fibre and provide the services. - At 08 September, 2009 11:06 Stephen Davies said...
-
Ger,
I would further add that the termination of 20,000 fibre plus the space required to house the OLTs plus the power systems to support all this is going to take many more racks. Again read some of the articles on my blog about this, and to service 5,000 homes you need about 125sqm of space using PtP. For 20,000 homes that is likely to be 400-500sqm of space. Unless you are a ILEC with lots of spare space in your exchanges this becomes an impactical proposition - At 08 September, 2009 17:13 Rudolf van der Berg said...
-
Hi Stephen,
Your understanding about Reggefiber working in conjunction with KPN using existing ducts is incorrect. Have a look at my blog. I have documented the building of the FTTH network to my home and you can see that everything is built new.
€1000 more than covers all that is necessary to make the network work for the end-user excluding the device that creates a CATV signal and Ethernet signal in the home and the active equipment the ISP brings in the home and the active equipment the ISP has in the neighbourhood's switching building.
Reggefiber carries all the costs of the network and all an ISP has to do is bring the active equipment. Regge's investment is below 1000 euro/household and it sells this for a regulated price of between 12 and 17 euro/month.
It is less clear for me what kind of costs the ISP has with all the customer premises equipment. In my OECD report I used a number from 2005 that was most likely too high at 760 euro, but it certainly wasn't too low. Almost all prices have come down in the last years, so hazard a guess of where we are now.
Reggefiber also provides services through its Lijbrandt and XMS subsidiaries. Other providers on its network are KPN, Concepts and Solcon.
All in all, for roughly 60 euro per month Dutch people have a triple play internet (50mbit), digital tv and telephony package.
Your very adamant that it has to be PON with a low fiber count and low split for consumers. All I'm trying to show you that the people who actually spend the money are choosing differently in some parts of the world.
My personal idea is that if you're correct about the lower TCO of PON, the choice should be for a PtP passive network topology and active GPON or WDM-PON equipment with a 32 way split. This way the end-user can burst up to 2.5Gbit/s, has a nice guaranteed amount of bandwidth and there is always the flexibility to switch to PtP Gigabit Ethernet or better if the end-user happens to work as a movie-editor or an MRI-technician. - At 08 September, 2009 18:34 Stephen Davies said...
-
Rudolf said...
"Your understanding about Reggefiber working in conjunction with KPN using existing ducts is incorrect."
Actually my understanding is quite correct. Much of the Reggefiber network does use the KPN duct infrastructure. However the access drops have tended to be new.
Your specific example of the new install fits that but is not the case everywhere.
Paraphrasing what you said above, the €1000 covers ONLY the outside plant. The ISP then provides all the equipment for the CATV signal, Ethernet home gateway, and the OLT in the neighborhood switching building.
So the cost is not €1000 for a complete end to end solution, it is much more when you include all the components that you require to deliver the services.
So back to the orginal 3-4 times the price difference, yes you can build a PON network for about €200. Opticomm (my employer) are doing that today using existing duct infrastructure for just the fibre component.
Then consider all the active equipment and again comparisons I have made only just recently support a 2-3 times price difference on equipment between PtP and PON - when considering ALL active equipment required to deliver a triple play of services.
Rudolf says...
"Your very adamant that it has to be PON with a low fiber count and low split for consumers."
No that is not correct. If you look at the presentation I refered to in the article I actually state Australia's strategy for the NBN requires a mix of both PtP and PON. PON for the residential play and PTP for high bandwidth subscribers.
I am not a PON bigot and I have deployed and operated both PON and PTP networks. I would suggest the company I work for would be one of a very few Carriers throughout the world which uses all three technologies (GPON, EPON and PtP) to deliver their services.
Its about selecting the right technology to fit the consumer requirements. As your blog points out you have 3 devices to deliver your three services. Using a PON technology you have 1 device to deliver those three services.
Rudolf says....
"All I'm trying to show you that the people who actually spend the money are choosing differently in some parts of the world."
I know that Europe is dominated predominately by PtP FTTP and there is nothing wrong with that - particularly with your very short loop lengths. There are two very good reasons why PtP is the predominant technology in Europe:
a) the regulatory environment about access to unbundled fibre;
b) Europe was one the first region worldwide to start deploying FTTH in the late 1990's and well before PON technology was around. So the use of existing Enterprise grade PtP Ethernet switches was the only option.
If you look around the world most of the major PtP FTTH vendors are European based companies.
Its also important to consider that Europe has a very different housing demography to Australia and the United states and our loop length are typically much longer - on average about 3.2kms.
With the rollout of FTTP one of the prime objectives of most incumbent and competitive carriers is to:
a) reduce active field equipment; and
b) reduce the number of exchanges which are expensive to run and maintain;
Rudolf said...
"All I'm trying to show you that the people who actually spend the money... "
While Europe is predominately rolling PtP, the rest of the world in predominately rolling PON. At current shipments of FTTH equipment, PON is about 82% of the market.
I might add that I am one of those poeple who are spending the money and have been for the past 9 years at various companies.