Bill, Jerry, Steve, et. al. I don't think I can over stress that the operational issue is not what the technology can support but what patterns the commercial fiber providers of the future will adopt. IMHO, operating separate channels of a WDM drop-and-insert will is no different from operating separate SONET/SDH drop-and-insert. Furthermore, the operational comparisions of multiple point-to-point links with a staticstically multiplexed link has been explored extensively on this list. However, can one *reasonably* expect that the owners of fiber in the ground to allow dark fiber access for the average (or even larger-than-average) network operator. Will the owners of the fiber allow the network operator to manipulate and re-structure his own drop-and-insert revisions over their dark fiber network? IMHO, the best you can reasonably expect is a dedicated port at both ends of a point-to-point link. If the price is right, you can expect to buy as many of these point-to-point links as you can afford. But physical control and network re-structuring will remain in the hands of the fiber plant owner-operator. Perhaps a network operator can plant dark fiber in a campus or regional deployment for extended WDM use. But I suspect the budget would need to be a lot larger (or you would need a really sweet political deal) to gain national access to dark fiber to implement and control the OADM or WDM equipment. The answers to Jerry's original question and the probability of a network operator truly controlling his own WDM network will come from future commercial practice of the fiber owners. ..mike.. At 09:31 AM 6/1/98 -0400, Bill St. Arnaud wrote:
Steven:
New municipal and campus WDM gear from Cambrian, Ciena and Nortel makes it a lot easier and cheaper to deliver data services over separate WDM channels than to MUX them into one SONET or ATM service. More importantly becuase the service is data transparent, each customer can manage their own WDM channel independent of the other channels. Try doing that with SONET or ATM!!
Check out www.cambrian.com for more details.
However, I agree with you that if you have spare fiber it is cheaper just to light up the fiber without WDM or SONET. Packet Engines claims they can drive 22km of dark fiber with no additional equipment or repeaters
Bill
------------------------------------------- Bill St Arnaud Director Network Projects CANARIE bill.st.arnaud@canarie.ca http://www.canarie.ca/bstarn
-----Original Message----- From: steven hessing [mailto:stevenh@inet.unisource.nl] Sent: Friday, May 29, 1998 5:08 PM To: Bill St. Arnaud Cc: Jerry Scharf; nanog@merit.edu Subject: Re: flat vs non-flat charging
Bill,
WDM may lower the price per Mbit/s considerably, however many of these Mbit/s will have to be sold to be profitable as a service. Connecting two routers using dedicated WDM gear and 5km of fiber to send over 155Mbit/s traffic is more expensive then connecting the two routers directly using 5km of fiber, even though the first solution would give you much more available capacity.
If multiple services need to run over the same fiber then existing ATM or SDH networks are well capable of multiplexing these services over one fiber at often lower costs then installing new WDM gear.
There where there is a fiber shortage because there is no further room for service multiplexing or there is a shortage of capacity, WDM will be a very good solution. It is much cheaper then installing new fiber.
WDM is therefor very real for connecting TEX's together and for connecting LEX's to TEX's. I expect that WDM needs OADMs or other up and coming technology for SDH or WDM to become cost-effective for the MAN/local loop.
To return to the original topic, for the forseeable future (3-5 years?) bandwidth will remain scarce and therefor expensive. The prices per MBit/s will drop shortply but this will be compensated in a large increase in demand. An distance-based pricing scheme could therefor be an appropriate solution. Personally, I don't think it will happen.
-- Steven
PS: if anyone can prove me wrong with some numbers, please show me :) I've been working on cost calculations for bandwidth for the last couple of weeks.
Jerry:
You are absolutely right. I have pulled together some numbers on public pricing for long haul WDM (including right of way) showing that if you ou= t IP over WDM the cost of bandwidth drops by a factor of 100 to 1000. On local loops it is more dramatic - with new WDM technology from companies like Cambrian and Ciena gigabit ethernet or OC-48 SONET on a 5km local lo= op should cost about $2500 per YEAR.
A number of our regional networks have already pulled their own fiber and our realizing these costs in the real world.
Fibre - $4 to $6 per meter 20 year economic life and 10% maintenance per year 48 strands NZDSF Right of way $0 -$10 per meter per year (up to $200 per meter on long hau= l) but common solution is to offer right way owner free use of a couple of strands instead of paying cash Installation $25 per meter underground in cities $6 per meter on poles (maintenance costs 20% year) Twenty year amortized cost with 10% cost of money plus maintenance plus right of way costs at $10/year per meter $17 per year per meter for 48 strands $1.50 per year per meter per strand $.50 per year per meter per wavelength A 5 km OC-48 local loop should cost about $2500/year
Bill
------------------------------------------- Bill St Arnaud Director Network Projects CANARIE bill.st.arnaud@canarie.ca http://www.canarie.ca/bstarn
=A0
=A0
-----Original Message----- From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu]On Behalf Of Jerry Scharf Sent: Friday, May 29, 1998 12:07 PM To: nanog@merit.edu Subject: flat vs non-flat charging
I think the idea of distance charging is going away in many cases. With WDM, the cost of the WDM and SONET eqiupment on the ends of a fully populated 32 channel per fiber, 144 strand pull vastly outweight end-to-end fiber costs of anything pulled through the ground. When you add routers and the like on top of that, the distance issue really goes away and it becomes on of netwo= rk topology hops. Can anyone with figures for new intercontinental pulls s= ay whether this is true there as well (project oxygen marketing claims thi= s, but...)?
Using archaic telephone pricing models to argue cost of
In your mail from 29-5-1998 you write: providing bulk =
IP
services is just not right.
jerry
Mike: I agree with you that a "customer" network operator would unlikely control the OADM components of a WDM network. That will remain in the domain of the fiber owner for the forseeable future. But, as per today's announcement with Microsoft and Nortel, the data transparency feature of regional or campus DWDM systems gives the "customer" network operator a lot more flexibility on the transport protocols on their point to point links. Billl ------------------------------------------- Bill St Arnaud Director Network Projects CANARIE bill.st.arnaud@canarie.ca http://www.canarie.ca/bstarn
-----Original Message----- From: Mike Trest (In Japan) [mailto:mike@trest.com] Sent: Wednesday, June 03, 1998 7:19 PM To: Bill St. Arnaud Cc: steven hessing; Jerry Scharf; nanog@merit.edu Subject: RE: flat vs non-flat charging
Bill, Jerry, Steve, et. al.
I don't think I can over stress that the operational issue is not what the technology can support but what patterns the commercial fiber providers of the future will adopt. IMHO, operating separate channels of a WDM drop-and-insert will is no different from operating separate SONET/SDH drop-and-insert. Furthermore, the operational comparisions of multiple point-to-point links with a staticstically multiplexed link has been explored extensively on this list.
However, can one *reasonably* expect that the owners of fiber in the ground to allow dark fiber access for the average (or even larger-than-average) network operator. Will the owners of the fiber allow the network operator to manipulate and re-structure his own drop-and-insert revisions over their dark fiber network? IMHO, the best you can reasonably expect is a dedicated port at both ends of a point-to-point link. If the price is right, you can expect to buy as many of these point-to-point links as you can afford. But physical control and network re-structuring will remain in the hands of the fiber plant owner-operator.
Perhaps a network operator can plant dark fiber in a campus or regional deployment for extended WDM use. But I suspect the budget would need to be a lot larger (or you would need a really sweet political deal) to gain national access to dark fiber to implement and control the OADM or WDM equipment.
The answers to Jerry's original question and the probability of a network operator truly controlling his own WDM network will come from future commercial practice of the fiber owners.
..mike..
Steven:
New municipal and campus WDM gear from Cambrian, Ciena and Nortel makes it a lot easier and cheaper to deliver data services over separate WDM channels than to MUX them into one SONET or ATM service. More importantly becuase the service is data transparent, each customer can manage their own WDM channel independent of the other channels. Try doing that with SONET or ATM!!
Check out www.cambrian.com for more details.
However, I agree with you that if you have spare fiber it is cheaper just to light up the fiber without WDM or SONET. Packet Engines claims they can drive 22km of dark fiber with no additional equipment or repeaters
Bill
------------------------------------------- Bill St Arnaud Director Network Projects CANARIE bill.st.arnaud@canarie.ca http://www.canarie.ca/bstarn
-----Original Message----- From: steven hessing [mailto:stevenh@inet.unisource.nl] Sent: Friday, May 29, 1998 5:08 PM To: Bill St. Arnaud Cc: Jerry Scharf; nanog@merit.edu Subject: Re: flat vs non-flat charging
Bill,
WDM may lower the price per Mbit/s considerably, however many of these Mbit/s will have to be sold to be profitable as a service. Connecting two routers using dedicated WDM gear and 5km of fiber to send over 155Mbit/s traffic is more expensive then connecting the two routers directly using 5km of fiber, even though the first solution would give you much more available capacity.
If multiple services need to run over the same fiber then existing ATM or SDH networks are well capable of multiplexing these services over one fiber at often lower costs then installing new WDM gear.
There where there is a fiber shortage because there is no further room for service multiplexing or there is a shortage of capacity, WDM will be a very good solution. It is much cheaper then installing new fiber.
WDM is therefor very real for connecting TEX's together and for connecting LEX's to TEX's. I expect that WDM needs OADMs or other up and coming technology for SDH or WDM to become cost-effective for the MAN/local loop.
To return to the original topic, for the forseeable future (3-5 years?) bandwidth will remain scarce and therefor expensive. The prices per MBit/s will drop shortply but this will be compensated in a large increase in demand. An distance-based pricing scheme could therefor be an appropriate solution. Personally, I don't think it will happen.
-- Steven
PS: if anyone can prove me wrong with some numbers, please show me :) I've been working on cost calculations for bandwidth for the last couple of weeks.
In your mail from 29-5-1998 you write:
Jerry:
You are absolutely right. I have pulled together some numbers on public pricing for long haul WDM (including right of way) showing that if you ou= t IP over WDM the cost of bandwidth drops by a factor of 100 to 1000. On local loops it is more dramatic - with new WDM technology from companies like Cambrian and Ciena gigabit ethernet or OC-48 SONET on a 5km local lo= op should cost about $2500 per YEAR.
A number of our regional networks have already pulled their own fiber and our realizing these costs in the real world.
Fibre - $4 to $6 per meter 20 year economic life and 10% maintenance per year 48 strands NZDSF Right of way $0 -$10 per meter per year (up to $200 per meter on long hau= l) but common solution is to offer right way owner free use of a couple of strands instead of paying cash Installation $25 per meter underground in cities $6 per meter on poles (maintenance costs 20% year) Twenty year amortized cost with 10% cost of money plus
At 09:31 AM 6/1/98 -0400, Bill St. Arnaud wrote: maintenance plus
right of way costs at $10/year per meter $17 per year per meter for 48 strands $1.50 per year per meter per strand $.50 per year per meter per wavelength A 5 km OC-48 local loop should cost about $2500/year
Bill
------------------------------------------- Bill St Arnaud Director Network Projects CANARIE bill.st.arnaud@canarie.ca http://www.canarie.ca/bstarn
=A0
=A0
-----Original Message----- From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu]On Behalf Of Jerry Scharf Sent: Friday, May 29, 1998 12:07 PM To: nanog@merit.edu Subject: flat vs non-flat charging
I think the idea of distance charging is going away in many cases. With WDM, the cost of the WDM and SONET eqiupment on the ends of a fully populated 32 channel per fiber, 144 strand pull vastly outweight end-to-end fiber costs of anything pulled through the ground. When you add routers and the like on top of that, the distance issue really goes away and it becomes on of netwo= rk topology hops. Can anyone with figures for new intercontinental pulls s= ay whether this is true there as well (project oxygen marketing claims thi= s, but...)?
Using archaic telephone pricing models to argue cost of providing bulk = IP services is just not right.
jerry
participants (2)
-
Bill St. Arnaud -
Mike Trest (In Japan)