RE: Weird GigE Media Converter Behavior
2 quick things come to mind- single mode vs multimode, or the signal is too hot and you need attenuators. -----Original Message----- From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu] Sent: Tuesday, August 24, 2004 12:48 PM To: nanog@nanog.org Subject: Weird GigE Media Converter Behavior Hello all, wondering if anyone has seen or experienced this same problem. Currently deploying an OC12 on a campus network and am using Netgear media converters. Set up the lines and had lit fiber to the building I needed it to go to. On one end (main drop) I had light and tested the connectivity which worked fine, on the other side, when I jacked up my Netgear, it flickered. Now, when I went to swap out the fiber, I noticed a full contact between the wires and the converter dropped it out. Meaning if I had the wires not fully plugged it lit, if I plugged them entirely in, it dropped. So... I ran to the other building and tried to replicate it there, and it did the same. I figured bad media converter so I swapped it out, next one worked fine. However, I took the media converter and swapped the funky acting one to yet another building and it worked fine. Anyone care to offer a thought on this? Relevant to my media converter? Bad line, "Murphy's Law", etc? =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+ J. Oquendo GPG Key ID 0x51F9D78D Fingerprint 2A48 BA18 1851 4C99 CA22 0619 DB63 F2F7 51F9 D78D http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x51F9D78D sil @ politrix . org http://www.politrix.org sil @ infiltrated . net http://www.infiltrated.net "How can we account for our present situation unless we believe that men high in this government are concerting to deliver us to disaster?" Joseph McCarthy "America's Retreat from Victory"
Yes, you have experienced what is known in the field as "air gap attenuation"--not something you want to depend on if you want a clean link! A close relative is "knot-in-the-fiber attenuation". Your subject says "GigE Media converter" but you say you are deploying an OC12. Which is it? Definitely invest in a light meter that can do the usual flavors of single and multimode, connector types, and at least 850 and 1310nm wavelengths. Then simply test strength of light on your receive port on each side, compare to specifications of your equipment, and add in-line attenuators as necessary. You'll usually find a range something like -3 to -27 dBm, and we prefer -15 dBm as our ideal. Also, make sure you've got the right type of fiber jumpers--mixing up single (yellow) and multi (orange) mode fiber can cause similar issues. Can't wait until more routers start to incorporate inline optical power readings in "show interface" commands the way Procket did :-) Cheers, -Lane On Tue, Aug 24, 2004 at 01:10:15PM -0400, kwallace@pcconnection.com <kwallace@pcconnection.com> wrote:
2 quick things come to mind-
single mode vs multimode, or the signal is too hot and you need attenuators.
-----Original Message----- From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu] Sent: Tuesday, August 24, 2004 12:48 PM To: nanog@nanog.org Subject: Weird GigE Media Converter Behavior
Hello all, wondering if anyone has seen or experienced this same problem. Currently deploying an OC12 on a campus network and am using Netgear media converters. Set up the lines and had lit fiber to the building I needed it to go to. On one end (main drop) I had light and tested the connectivity which worked fine, on the other side, when I jacked up my Netgear, it flickered. Now, when I went to swap out the fiber, I noticed a full contact between the wires and the converter dropped it out. Meaning if I had the wires not fully plugged it lit, if I plugged them entirely in, it dropped.
So... I ran to the other building and tried to replicate it there, and it did the same. I figured bad media converter so I swapped it out, next one worked fine. However, I took the media converter and swapped the funky acting one to yet another building and it worked fine. Anyone care to offer a thought on this? Relevant to my media converter? Bad line, "Murphy's Law", etc?
=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+ J. Oquendo GPG Key ID 0x51F9D78D Fingerprint 2A48 BA18 1851 4C99
CA22 0619 DB63 F2F7 51F9 D78D http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x51F9D78D
sil @ politrix . org http://www.politrix.org sil @ infiltrated . net http://www.infiltrated.net
"How can we account for our present situation unless we believe that men high in this government are concerting to deliver us to disaster?" Joseph McCarthy "America's Retreat from Victory"
On Tue, 24 Aug 2004, Deepak Jain wrote:
Can't wait until more routers start to incorporate inline optical power readings in "show interface" commands the way Procket did :-)
Don't SFPs provide this sort of optical digital diagnostics?
Apparently the CRS-1 supports this, as well as a few other types of GBIC's.
* lane@laneandmimi.com (Lane Patterson) writes:
Can't wait until more routers start to incorporate inline optical power readings in "show interface" commands the way Procket did :-)
* deepak@ai.net (Deepak Jain) adds:
Don't SFPs provide this sort of optical digital diagnostics?
Yes, but hardly any vendor provides ways to extract this highly useful information from the hardware. I think the Cat6.5k does. The Foundry MG8 shows you the serial #'s of inserted SFPs and XENPAKs in `show media'; useful in a different way. -- Niels.
Also, some of the SFP vendors did not necessarily provide all the optical power reading, transmit bias current, voltage, temperature etc few years ago. Nowadays it is more common for the SFP vendors to provide this information (and in a standard fashion) enabling system vendors to be more "bold" about displaying these parameters from their CLI/management systems. Vinay Bannai Luminous Networks ----- Original Message ----- From: "Niels Bakker" <niels=nanog@bakker.net> To: <nanog@merit.edu> Sent: Tuesday, August 24, 2004 3:23 PM Subject: Re: Weird GigE Media Converter Behavior
* lane@laneandmimi.com (Lane Patterson) writes:
Can't wait until more routers start to incorporate inline optical power readings in "show interface" commands the way Procket did :-)
* deepak@ai.net (Deepak Jain) adds:
Don't SFPs provide this sort of optical digital diagnostics?
Yes, but hardly any vendor provides ways to extract this highly useful information from the hardware.
I think the Cat6.5k does. The Foundry MG8 shows you the serial #'s of inserted SFPs and XENPAKs in `show media'; useful in a different way.
-- Niels.
At 01:31 PM 8/24/2004 -0700, Lane Patterson wrote:
Yes, you have experienced what is known in the field as "air gap attenuation"--not something you want to depend on if you want a clean link! A close relative is "knot-in-the-fiber attenuation".
a close relative of "who-pulled-this-fiber-anyway" ?" yes .. a kink ..a slight kink can do much loss .. easy to hold a fiber around your finger w/a meter attached and watch the atten .. as you bend the fiber ..
Your subject says "GigE Media converter" but you say you are deploying an OC12. Which is it?
I wondered that also ..
Definitely invest in a light meter that can do the usual flavors of single and multimode, connector types, and at least 850 and 1310nm wavelengths. Then simply test strength of light on your receive port
I'd go w/a test set if you afford it. It'll serve more than purpose and you'll get more use of it ..
on each side, compare to specifications of your equipment, and add in-line attenuators as necessary. You'll usually find a range something like -3 to -27 dBm, and we prefer -15 dBm as our ideal. Also, make sure you've got the right type of fiber jumpers--mixing up single (yellow) and multi (orange) mode fiber can cause similar issues.
Can't wait until more routers start to incorporate inline optical power readings in "show interface" commands the way Procket did :-)
some flava's of the GSR will show/report this .. under show controller but cisco says it's +_ 5 dbm .. they say .. ' You want a router or a meter ? " I have to agree .. would want not the expense of this added in. POS3/0 SECTION LOF = 0 LOS = 0 BIP(B1) = 0 LINE AIS = 0 RDI = 0 FEBE = 0 BIP(B2) = 0 PATH AIS = 5 RDI = 6 FEBE = 387 BIP(B3) = 6389 LOP = 5 NEWPTR = 0 PSE = 0 NSE = 0 Framing: SONET APS Optical Power Monitoring Rx optical power in mWs and dBms Port 0 = 0.02 mW, - 15.738 dBm Tx laser diode forward bias current I(F) in milliamps Port 0 = 18.009 mA Clock source: line
Cheers, -Lane
On Tue, Aug 24, 2004 at 01:10:15PM -0400, kwallace@pcconnection.com <kwallace@pcconnection.com> wrote:
2 quick things come to mind-
single mode vs multimode, or the signal is too hot and you need
attenuators.
-----Original Message----- From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu] Sent: Tuesday, August 24, 2004 12:48 PM To: nanog@nanog.org Subject: Weird GigE Media Converter Behavior
Hello all, wondering if anyone has seen or experienced this same problem. Currently deploying an OC12 on a campus network and am using Netgear media
On Tue, 24 Aug 2004, John R. Sosebee wrote:
I have to agree .. would want not the expense of this added in.
At the huge markup Cisco does regarding their optics, they have the margin to do this without hiking the price. They're already pricing in a complete SDH system in their optics markup. I noticed that the CRS-1 is now in the Cisco pricelist. 4 port OC192 VSR $330k 4 port OC192 SR $630k 4 port OC192 IR $1030k I happen to know a 10GE SR Xenpak can be had at $1600 and LR (80km) at $5900 from manufacturer. The same laser can do OC192. Is there anyone who can justify this pricing with anything else than "because we can?" -- Mikael Abrahamsson email: swmike@swm.pp.se
Mikael Abrahamsson <swmike@swm.pp.se> writes:
Is there anyone who can justify this pricing with anything else than "because we can?"
To expand on what I said to you privately, let's follow the money: Assume $200,000/board as the marginal cost of manufacturing one. Assume a minimum of 65 points off for any customer who buys a CRS-1 (hint: large telcos do not pay list price... and nobody, not even certain well-known crazy people, is going to pay money for one to use as a SOHO router). So now we have a strawman marginal profit per board sold of $150,000. Assume that Cisco actually has about a billion dollars into the HFR project (offhand comment by a usually reliable source). That means that ignoring the chassis sales side, we're looking at about 6700 interface cards (> 26500) OC192 IR ports sold to get back the initial engineering investment. That's not figuring in the cost of support for those 6700 cards out in the field; you're probably looking at 8500 or more cards (> 34000 ports) sold in order to hit break-even. Now consider the size of the market. You're not selling PAs for people's 7200s here, and you're selling a card that replaces *four* cards on a lesser platform. Under the circumstances, I'm not sure that Cisco's list price is high enough, but I'll defer judgement on that to the people who run the numbers behind the scenes in San Jose. They're the ones whose jobs are on the line if this product turns out to be $1b bragging rights exercise with no hope for payback. If you find the prices staggering, it's likely that you and your organization don't need this product. Arguments about price gouging on memory, GBICs, power cords, and other commodity items that your organization actually *does* need are orthogonal to this discussion. ---Rob
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 2004-08-29, at 15.58, Robert E. Seastrom wrote:
If you find the prices staggering, it's likely that you and your organization don't need this product. Arguments about price gouging on memory, GBICs, power cords, and other commodity items that your organization actually *does* need are orthogonal to this discussion.
didn't we have this discussion when the T640 came out. How many have one? - - kurtis - -----BEGIN PGP SIGNATURE----- Version: PGP 8.1 iQA/AwUBQTWI9qarNKXTPFCVEQIXDQCcCBnI+QMGSbojRDqGi1rN854/AHEAoLFY jusxejesCvB+3OA4AImajpmO =CVTQ -----END PGP SIGNATURE-----
--On onsdag 1 september 2004 10.31 +0200 Kurt Erik Lindqvist <kurtis@kurtis.pp.se> wrote:
didn't we have this discussion when the T640 came out. How many have one?
Nordunet has one. Nice box. -- Måns Nilsson Systems Specialist +46 70 681 7204 KTHNOC MN1334-RIPE
On Tue, Aug 24, 2004 at 10:45:22PM -0400, John R. Sosebee <sosebee@bellsouth.net> wrote:
some flava's of the GSR will show/report this .. under show controller but cisco says it's +_ 5 dbm .. they say .. ' You want a router or a meter ? " I have to agree .. would want not the expense of this added in.
Cool, upon further discussion with hardware guys, most of this functionality is being built into the 3rd party optical components these days, so all the vendor has to do is augment the CLI to pass this data through. If the capability is there, why not use it? I must say it would be cool to graph dBm over months/years in Cricket, and see if you could spot fiber degradation. However, as you point out, it is important for vendors to document the accuracy of the readings in their spec sheets. +/-5 dBm sounds a bit lame. As I understand, these components split off about 2% of the light to take the optical power readings. Don't know if any of them can measure reflection to get distance as well? -Lane
POS3/0 SECTION LOF = 0 LOS = 0 BIP(B1) = 0 LINE AIS = 0 RDI = 0 FEBE = 0 BIP(B2) = 0 PATH AIS = 5 RDI = 6 FEBE = 387 BIP(B3) = 6389 LOP = 5 NEWPTR = 0 PSE = 0 NSE = 0
Framing: SONET APS
Optical Power Monitoring Rx optical power in mWs and dBms Port 0 = 0.02 mW, - 15.738 dBm
Tx laser diode forward bias current I(F) in milliamps Port 0 = 18.009 mA
Clock source: line
On Wed, 25 Aug 2004, Lane Patterson wrote:
the optical power readings. Don't know if any of them can measure reflection to get distance as well?
There are even some with simple OTDR functionality built into them, just like there are some copper ethernet PHYs that also have this (CTDR). <http://www.opticalzonu.com/products/P2P%20Gig%20Ethernet%20OTDR%20Transceivers.html> -- Mikael Abrahamsson email: swmike@swm.pp.se
participants (11)
-
Deepak Jain -
John R. Sosebee -
Kurt Erik Lindqvist -
kwallace＠pcconnection.com -
Lane Patterson -
Mikael Abrahamsson -
Måns Nilsson -
Niels Bakker -
Robert E. Seastrom -
Tom (UnitedLayer) -
Vinay Bannai