This presentation is one of mine which was delivered by a work colleague at the recently Paul Budde FTTH round table. It raises a number of technology issues in regards to the proposed National Broadband Network and which technology is best suited to the deployment.
This presentation has been born out of a very frustating arguement I had with several members of Whirlpool who were addement PtP is the ONLY technology which should be considered, and were coming up with some very silly arguements such as shining lasers down the fibre.
The five main myths I dispell are:
1. PtP offers 100Mbps to each sub, but PON does not because it shares bandwidth
2. PtP offers easy upgrade path, PON does not
3. PtP is cheaper because it uses cheaper optics
4. PtP has a dominant market share
5. PON is insecure and prone to DoS attacks
I hope you enjoy the presentation.
- At 30 May, 2009 05:33 Zed Usser said...
I do believe your presentation to be a collection of unasserted opinions and factual fallacies. Furthermore, it is for the worse that the presentation neglects to address the most pertinent key issues in the comparison between PON and PtP.
- Option value: a PtP topology has a far larger inherent option value than a PON topology. Any serious comparison should at least aspire to compare option value costs and replacement costs for lost option value.
- Asymmetry: Current PON solutions can not offer symmetric bandwidth at high split ratios
- Interoperability: PON vendors have yet to show device interoperability unlike most PtP vendors
Some of the presented positions might be of merit, but the supporting argumentation is so dismal that it gives the whole presentation no more credibility than the uninformed quips of your vocal opponents.
I do hope you either update the presentation with verified facts, corrections and qualified references or remove it, as per your self-proclaimed web policy. Anything other would be a great disservice to the industry.
The following list details the matters that most require redress in the presentation in addition the above deficiencies.
Page 7 - Point to point: Citation on cost analysis needed
Page 11 - Dispelling the Myths: Citation on topology analysis needed
Page 12 - Myth 1: Inaccurate product comparison, failure to address asymmetry of PON solution
Page 13 - Myth 2: Failure to recognize that no PON ONT may be upgraded without OLT forklift upgrade, failure to assess viability of upgrade path in light of non-existent PON vendor interoperability, failure to recognize that non-implemented and non-standardized projected PON upgrade path is not an actual and actionable upgrade path
Page 14 - Myth 3: Citation on CAPEX and TCO analysis needed
Page 15 - Myth 4: Citation on market share data needed
Page 16 - Myth 5: Citation on PON DoS research needed
Page 17 - Myth 6: Citation on engineering feasibility study needed, failure to differentiate between ULL and bitstream
Page 20 - Residential Areas: Failure to differentiate between absolute truth and personal recommendation
Page 22 - Open Access: Failure to recognize that bitstream is a subset of LLU
- At 30 May, 2009 10:15 Stephen Davies said...
Well obviously you work for a PtP Vendor!
I have been designing both PtP and PON fTTH solutions for more than 8 years and have worked with many different vendor solutions, so I think my opinion is well founded on evidence and experience of real life deployments.
All you have done is thrown up more myths that are further perpetuated by the vendors of PtP technology.
You demand that I provide citations of my own research and work, but you dont even identify yourself or your statements.
PtP has a far larger inherent option value than PON. Based on what? PON can deliver any service that PtP can in a residential access network, further more it can deliver it more efficently.
Asymmetry: PON networks most certainly offer symmetric bandwidth services at typical split ratios of up to 64:1. All you are trying to suggest here is the same myth about bandwidth. For residential access networks, most subscribers do not receive a 1:1 contention and bandwidth is shared extensively throughout the network. It make no difference if it is PtP or PON.
Interoperability: This arguement is completely meaningless and another myth by the PtP vendors. How often do carriers mix technologies in the same access network? Extremely rarely! So what is the point or need for interop? At the network layer (IP) the equipment will interop with any other network, at the datalink layer you want a tightly coupled end-to-end solution based on a single vendor.
In the PtP world there are few vendors who offer a tightly coupled solution which includes OLT and ONT (customer premises device) that can be managed through a single element management platform. This is key to a large scale cost effective service delivery and management.
Page 7. Read some of the white papers I have produced. There is detailed cost analysis in them based on my evaluations of different FTTH access solutions. The white paper is a summary of the detailed tender and analysis which was conducted previously for one of my clients and more recently for the company which I now work for.
Page 12. Asymmetry in the access has nothing to do with symmetry of services delivered to the subscriber. Besides, residential access networks are typically asymmetrical in nature.
Also your citation for this myth? NTT is Japan is delivering 100Mbps symmetrical services over its EPON network.
Here in Australia Opticomm and Nextgen Networks are delivering 100Mbps symmetrical services over a PON network.
Page 13. Again a complete myth. A well design fibre plant can address the ability to upgrade OLTs and ONT independantly of each other, while still using a PON topology.
Furthermore, what upgrade is needed at this point in time? To 10GE PON? Well the design of 10GE PON enables both 1G PON or 2.5GPON to interoperate at the same time on the same PON with 10GEPON, without forklift upgrade.
As for PtP? To upgrade from 100Mbps access to 1Gbps access does require replacement of the OLT (switch) and ONT at the same time, or repatching of cables.
Myth 3. It my own work and experience of deploying FTTH networks in Australia and overseas for the past 8 years. Its also well documented by Verizon and NTT.
Myth 4. As above, the information is mased on the analysis I have performed product evaluations through tenders.
Myth 5. Experience and testing in our own Access networks.
Myth 6. ULL in a fibre network is completely pointless. Open Access at the bitstream layer is the most cost effective mechanism of service delivery. Citation Yankee Group which is one of the presentations on this web site.
- At 30 May, 2009 10:15 Stephen Davies said...
Page 20. Experence in network deployments. Citation given of the two largest FTTH deployments in the world being PON based. As for the deployment capability and scaling, please name one PtP deployment which is supporting hundreds of thousands or millions of subscribers TODAY. That is DEPLOYED not going to be DEPLOYED.
Furthermore there are articles on this site which explain the scaling issues of PtP.
Page 22. Agreed that bitstream is a subset of LLU, but is far more effective (particularly in cost) in the delivery of services than ULL. You need to come into the 21st century in how telecommunications services are going to be delivered. The arguement of ULL access to the encumbant carrier's network is a thing of the past. Wholesale bitstream access is far more effective in the delivery of services than ULL.
- At 30 May, 2009 10:17 Stephen Davies said...
I would more than welcome an online public debate on this matter with you.
I have all the evidence to support my statements; I can assure you that the PtP vendors do not.
- At 02 June, 2009 05:05 John Nixon said...
Good on your Stephen, I'm with you!
Keep up the good work.
Optical Network Engineering
Auckland - New Zealand
- At 02 June, 2009 14:13 Hamish said...
Please forgive the question if it is naive, but you're very certain about the advantage open access at the bitstream holds over optical ULL. I'd appreciate your reasoning behind the claim of "more effective and efficient open access at the bitstream?"
With PON and bitstream "open access" I see only the perpetuation of a single closed last mile, where retailers have a monopoly wholesaler to make decisions about what bitstreams to offer.
That recapture might make your views on PON's advantages moot.
- At 02 June, 2009 15:32 Stephen Davies said...
The advantages bitstream access has over Fibre ULL is the issue of cost of network deployment and maintenance. If you are building a national broadband network, you are going to connect every home as you pass them by. This considerably reduces the cost of customer connection.
If you provide only a fibre ULL service, then each home has to be visited to install the ONT and establish the connection. When that customer moves house, or wants to change provider, they need to change the ONT.
ONTs are not CPE and should not be treated like ADSL routers.
Furthermore, how do you control the standardisation of ONT offering to ensure you can support other services such as smart grid, eHealth etc. without the use of the internet.
Some sugguest SmartGrid should be provided using a seperate fibre and equipment, but this adds further cost to deployment and maintenance of the network.
- At 03 June, 2009 05:42 Zed Usser said...
I think you have your wires crossed, Steven. ONTs aren't CPEs...
(i) ULL does not require truck rolls nor does ULL require the customer to change or move ONTs
(ii) If you are hooking up everybody during rollout, it really does not matter whether you are doing bitstream or ULL for up front costs
(iii) ULL need not be more expensive than bitstream
Where you are right on the money is standardisation. The goal is obviously standardised CPEs, competition, large manufacturing volumes and self installs.
- At 03 June, 2009 05:47 Zed Usser said...
Hamish: Right on!
For more details on why bitstream access only is to be considered harmfull, see my comments on Steven's other post on open access: http://ozftth.blogspot.com/2009/06/bitstream-or-ull-open-access.html
- At 03 June, 2009 05:50 Zed Usser said...
Steven: I'll get back to you on the other stuff later on once I've some spare time.
- At 03 June, 2009 09:34 Stephen Davies said...
Zed Usser said "I think you have your wires crossed, Steven. ONTs aren't CPEs..."
I think you have your wires crossed, I never said ONTs are CPE, quite the opposite.
My exact statement "ONTs are not CPE and should not be treated like ADSL routers."
Zed Usser said "ULL does not require truck rolls nor does ULL require the customer to change or move ONTs"
So if ULL is being provided by a wholesale carrier for access by multiple service providers, who is installing the ONT? The customer? I dont think so. So how is the ONT installed without a trunk roll? Its not possible.
"If you are hooking up everybody during rollout, it really does not matter whether you are doing bitstream or ULL for up front costs"
You are making a comparsion of cost between a layer 3 service and a layer 2 service which is not practical. However if you compare a ULL service which requires a PtP topology to a PON (which can deliver bitstream) there is a significantly higher cost for PtP fibre over PON.
- At 09 October, 2009 21:16 myne said...
I argue for a PtP cabling layout with PON equipment in the exchange.
PtP cabling takes the guesswork out of the future.
PON equipment takes the dollar cost out of the initial equipment rollout.
Over time, should it be demanded, then you can optionally chose to use ULL/PtP layout.
Think of it this way. Currently with ADSL over the copper CAN we have a physical PtP.
However, 12-32 pairs connect to a single DSLAM card. If that DSLAM card was to be set to broadcast with encryption and time division, like the HFC and PON networks do it would approximate what I am suggesting. Please dont nitpick on how impossible this is with ADSL. It's a concept, not a plausible solution.
Physical PtP for flexibility. PON equipment for the baseline.
One obvious advantage of this is congestion. If one segment is over subscribed, or the users are particularly demanding, you can switch some customers to a different PON segment to even out the contention.
Migrations to newer technologies can be staged, in the instance that they are not backwards compatible.
You can shift users one by one onto the newer cards at the exchange end.
Over time, you would expect that newer technologies make the base hardware redundant. Well, not entirely. In much the same way we still have a lot of people on dialup or legacy ADSL hardware, you can incrementally upgrade the customers, thus leaving those who dont require the latest and greatest to enjoy reduced cost benefits of being connected to shared, fully paid off equipment.
At the end of the day, it's likely that even the least efficient PON of today, once paid off in full, is still going to be quite economical electrically and CAPEX wise than a newer card.
As for the comments on ULL vs wholesale bitstream, I see many advantages to a wholesale bitstream service.
For starters, with a fibre ULL, it's going to be a mighty headache to have your voice and data services with differnt providers.
I like the potential for a wholesale bitstream service to have multiple services billed by different companies travelling down the same transmissions medium.
Free TV, Pay TV, telephony, internet, Free .gov sites only internet.
These are just some examples.
Going further, you might be able to pay one company for your internet, and another a subscription to their gaming network only without internet access. I'd assume you'd do this with vlan tags.
This of course means that the network must support QoS and all CPE must be capable of decent routing.
That's in an ideal world I admit.
PtP has its merits. It's definitely a much better cabling option in my inexperienced opinion.
PON has its merits. It's a pragmatic solution that gets the damn job done for a reasonable price.
Zed Usser, will you concede that what you really care about is the flexibility of the network and not the particular choice of transmission?
You want the OPTION of PtP. You dont want to be locked into a potentially limiting technology.
You dont really care about the other arguements.