Good Evening all. I got an odd and somewhat crazy request from our development group for a long haul OC48 connection for testing (they specifically said from their office in Utah to the east coast and back) with minimal jitter. They need to be able to run their own framing Sonet and WDM - don't ask me why, it doesn't make sense to me. It would seem to me that the last requirement would require a dark fiber? So I've got several questions. One, is there any provider that would provide us such a beast for only one month? I realize that the fact this is long haul OC48 would make the cost astronomically high, and then throwing on the one month would make it even more expensive.... Two, is there any good way to simulate such a long distance link? I know such equipment exists for smaller links, but I haven't yet found anything that'll do OC48 fiber. I'm sure the cost would be high for the equipment, however I'm betting it is *much* lower than the long distance link. Three, could we sanely encapsulate their framing into GRE (or some other such technology) and send it over an IP link and get SLAs to minimize the jitter? Offlist responses would be greatly appreciated.
On 3/10/2011 12:15 AM, nanog wrote:
Good Evening all. I got an odd and somewhat crazy request from our development group for a long haul OC48 connection for testing (they specifically said from their office in Utah to the east coast and back) with minimal jitter. They need to be able to run their own framing Sonet and WDM - don't ask me why, it doesn't make sense to me. It would seem to me that the last requirement would require a dark fiber?
So I've got several questions.
One, is there any provider that would provide us such a beast for only one month? I realize that the fact this is long haul OC48 would make the cost astronomically high, and then throwing on the one month would make it even more expensive....
Two, is there any good way to simulate such a long distance link? I know such equipment exists for smaller links, but I haven't yet found anything that'll do OC48 fiber. I'm sure the cost would be high for the equipment, however I'm betting it is *much* lower than the long distance link.
Three, could we sanely encapsulate their framing into GRE (or some other such technology) and send it over an IP link and get SLAs to minimize the jitter?
Offlist responses would be greatly appreciated.
Would it be too crazy to buy a spool of fiber and splice the end of one pair to the next pair and so on? Won't be able to simulate 2200 miles of fiber but it'll be a long span.
On Fri, Mar 11, 2011 at 9:25 AM, ML <ml@kenweb.org> wrote:
Would it be too crazy to buy a spool of fiber and splice the end of one pair to the next pair and so on? Won't be able to simulate 2200 miles of fiber but it'll be a long span.
This is by no means crazy. If you visit a laboratory where gear is tested, you'll find exactly that -- spools of fiber which can be connected together (through whatever splicing or patching method is desired for the simulation) to give the desired span length. These usually look nicer than big spools of cable, and are even available in rack-mount enclosures with vendor logos. :) -- Jeff S Wheeler <jsw@inconcepts.biz> Sr Network Operator / Innovative Network Concepts
On 3/11/11 7:16 AM, Jeff Wheeler wrote:
On Fri, Mar 11, 2011 at 9:25 AM, ML <ml@kenweb.org> wrote:
Would it be too crazy to buy a spool of fiber and splice the end of one pair to the next pair and so on? Won't be able to simulate 2200 miles of fiber but it'll be a long span.
This is by no means crazy. If you visit a laboratory where gear is tested, you'll find exactly that -- spools of fiber which can be connected together (through whatever splicing or patching method is desired for the simulation) to give the desired span length. These usually look nicer than big spools of cable, and are even available in rack-mount enclosures with vendor logos. :)
one does not however do 2200 miles of terrestrial fiber simulation without simulating regeneration as well.
On Mar 11, 2011, at 7:50 AM, Joel Jaeggli wrote:
On 3/11/11 7:16 AM, Jeff Wheeler wrote:
On Fri, Mar 11, 2011 at 9:25 AM, ML <ml@kenweb.org> wrote:
Would it be too crazy to buy a spool of fiber and splice the end of one pair to the next pair and so on? Won't be able to simulate 2200 miles of fiber but it'll be a long span.
This is by no means crazy. If you visit a laboratory where gear is tested, you'll find exactly that -- spools of fiber which can be connected together (through whatever splicing or patching method is desired for the simulation) to give the desired span length. These usually look nicer than big spools of cable, and are even available in rack-mount enclosures with vendor logos. :)
one does not however do 2200 miles of terrestrial fiber simulation without simulating regeneration as well.
You can, but, it requires electronic retiming of the fiber signal (fiber->ring buffer->configurable delay->fiber). I guess technically that simulates one iteration of regeneration to some extent, but, it certainly wouldn't represent a test of 2200 miles worth of analog regeneration of a digital signal. Owen
participants (5)
-
Jeff Wheeler -
Joel Jaeggli -
ML -
nanog -
Owen DeLong