DataCenter color-coding cabling schema
Hello Nanog-ers, Have any of you had the option or; conversely, do you know of “best practices" or “common standards”, to color code physical cabling for your connections in DataCenters for Base-T and FX connections? If so, Could you share any ttype of color-coding schema you are aware of ?…. Yes, this is actually considering paying for customized color-coded cabling in a Data Center... Mr. Google did not really provide me with relevant answers on the above… beyond the typical (Orange is for MMF, yellow for SMF, etc)… Any reasons for or against it welcome too... -- Yardiel Fuentes
I know at Clearwire data centers we used gray for network, blue for management and orange for RS-232 console. At least for the initial build. Later re-work or additions were whatever the tech had on hand ;) They also had labels on each end of each wire showing the path through the system, sometimes up to six lines. It did make it easy to bring up a data center and find cabling errors. To see the system last more than a year or two up upgrades would take some strong rules and oversight. I think it would be worth it if your management system can keep the religion. -- Joe Hamelin, W7COM, Tulalip, WA, +1 (360) 474-7474 On Sat, Mar 12, 2016 at 11:11 AM, Yardiel Fuentes <yardiel@gmail.com> wrote:
Hello Nanog-ers,
Have any of you had the option or; conversely, do you know of “best practices" or “common standards”, to color code physical cabling for your connections in DataCenters for Base-T and FX connections? If so, Could you share any ttype of color-coding schema you are aware of ?…. Yes, this is actually considering paying for customized color-coded cabling in a Data Center...
Mr. Google did not really provide me with relevant answers on the above… beyond the typical (Orange is for MMF, yellow for SMF, etc)…
Any reasons for or against it welcome too...
-- Yardiel Fuentes
On 3/12/16 12:15 PM, Joe Hamelin wrote:
I know at Clearwire data centers we used gray for network, blue for management and orange for RS-232 console. At least for the initial build. Later re-work or additions were whatever the tech had on hand ;) They also had labels on each end of each wire showing the path through the system, sometimes up to six lines. It did make it easy to bring up a data center and find cabling errors. To see the system last more than a year or two up upgrades would take some strong rules and oversight. I think it would be worth it if your management system can keep the religion.
That's the issue, keeping it that way. "Gray for network" is likely to result in mostly gray cables which won't really help to differentiate things in the long run. Breaking it down further can get tricky in terms of definition. Each network has a color, but then there's this trunk link.... We had a customer who had a scheme involving five different colors. When they did the initial build their wiring vendor came in with barrels of new cables of various lengths and colors and it looked really nice with cable management and all. After a couple of years it was pretty much random in terms of color coding. Keeping multiple lengths on hand for dressing in raceway without incurring either tons of slack or bow-string taut wires is tough but possible, doing that in half a dozen colors can be daunting. The electrons on the inside can't see the jacket on the outside and most of them are rumored to be color-blind anyway. Maybe a compromise of a single color for most things and a different one for specials. -- Jay Hennigan - CCIE #7880 - Network Engineering - jay@impulse.net Impulse Internet Service - http://www.impulse.net/ Your local telephone and internet company - 805 884-6323 - WB6RDV
On Mar 18, 2016, at 18:42 , Jay Hennigan <jay@west.net> wrote:
On 3/12/16 12:15 PM, Joe Hamelin wrote:
I know at Clearwire data centers we used gray for network, blue for management and orange for RS-232 console. At least for the initial build. Later re-work or additions were whatever the tech had on hand ;) They also had labels on each end of each wire showing the path through the system, sometimes up to six lines. It did make it easy to bring up a data center and find cabling errors. To see the system last more than a year or two up upgrades would take some strong rules and oversight. I think it would be worth it if your management system can keep the religion.
That's the issue, keeping it that way. "Gray for network" is likely to result in mostly gray cables which won't really help to differentiate things in the long run. Breaking it down further can get tricky in terms of definition. Each network has a color, but then there's this trunk link....
We had a customer who had a scheme involving five different colors. When they did the initial build their wiring vendor came in with barrels of new cables of various lengths and colors and it looked really nice with cable management and all.
After a couple of years it was pretty much random in terms of color coding. Keeping multiple lengths on hand for dressing in raceway without incurring either tons of slack or bow-string taut wires is tough but possible, doing that in half a dozen colors can be daunting.
Yes and no. If you have a requirement that all cabling between racks goes via fixed cabling from patch panels and patch cables are only used for intra-rack runs between equipment in the same rack and/or equipment<->patch panel in the same rack, it gets a lot less so. This can also help keep a lot of other things more sane in the long run as well. I found that you could deal pretty well with any intra-rack run (assuming 7’ racks) if you stocked the following lengths: 0.5’ 1’ 1.5’ 2’ 2.5’ 3’ 4’ 5’ 6’ 7’ 8’ 10’ That’s a total of 12 lengths. We kept those in stock in Yellow (SMF), Orange (MMF), Other colors all Cat 6: Blue, Red, Purple, Green. Total of 72 part numbers to keep track of. We got one of those roll-around bin carts that had 4 rows of 9 drawers on each side. Worked out perfectly to have 72 kinds of cables. (IIRC, we did 3 columns per color working up in size from left to right). We had a guy who was responsible for making sure none of the bins were ever empty. I think he checked the cart twice a week and ordered once a month most months. I think we tended to keep at least 5 on the cart and topped the bins up to 20 for most sizes and 40 for the popular size/color combinations. We didn’t have any trouble maintaining the system and as long as the right color was available, we got pretty good compliance from the people installing cables. (Our “retraining” method for people who ran a wrong-colored cable didn’t hurt, either.) YMMV. Owen
I don’t know of any universal standards, but I’ve used the following in several installatins I was responsible for to good avail: Twisted Pair: RED: Untrusted Network (Internet or possibly DMZ) YELLOW: Optional for DMZ networks though I preferred to avoid documented in [1] below BLUE: Trusted Network (back-end, internal, etc.) GREEN: RS-232 straight-thru PURPLE: RS-232 X-Over (effectively Null Modem) 12345678 <-> 87654321 pin map. ORANGE: Ethernet X-Over (Best avoided documented in [2] below) GREY: Special purpose cabling not in one of the above categories Fiber: Orange — Multimode Fiber Yellow — Singlemode Fiber The absolute most useful thing you can do if you can impose the discipline to update the cable map rigorously and/or allocate manpower for periodic audits is to apply a unique serial number to each cable. I preferred to document not only the cable ID, but also the length. For the installations where I have worked, 5 digits was sufficient unique ID, so I used formats like IIIII-L[.L] where IIIII was a unique ID and L.L was the length of the cable in feet. (e.g. 00123-6.5 is cable number 123 which is 6.5 feet in length). The labels are (ideally) the self-laminating wrap-around types. I prefer the Brady labeling system which will automatically print 2-4 (depending on font size) instances of the label text on the self-laminating label such that it can be read from virtually any side of the cable without requiring you to rotate the label into view in most cases. The Brady labeling system is a bit overpriced compared to the Brother P-Touch, but the expanded capabilities and the quality of the label adhesives and such is, IMHO, sufficiently superior to justify the cost. Whatever you do, please do not use Flag labels on cables… I HATE THEM. They are a constant source of entanglement and snags. They often get knocked off as a result or mangled beyond recognition, rendering them useless. Similarly, I’ve found that circuit-ID and end-point labels on cables are often ill-maintained, so if you do use them, please make sure you remove them when the cable is moved/removed. The length is very useful because it gives you a radius within which the other end of the cable must be located and you can usually expect it to be reasonably close to the outer edge of that radius. More than a few times I’ve prevented a serious outage by giving the port number to the remote hands guy and then insisting that he read me the cable ID. “No, try the other port FE-0/2/4… You’re off by one. It’s above/left/right/below you.” [1] I prefer to avoid Yellow cables because some people have trouble understanding that Yellow Fiber and Yellow UTP might have different meanings. I also feel that the distinction between UNTRUSTED and DMZ networks is usually not all that important in most cabling situations. YMMV. [2] In this era of Auto-MDI/MDI-X ports and the like, it’s very rare to encounter a situation that truly requires a crossover cable with no viable alternative. If such is needed, I prefer to document it on the cable tags rather than using a special color code. Again, you have the risk of people not understanding that orange Fiber might not mean what Orange copper means. YMMV Yes, I know you can now get virtually any type of fiber in virtually any color, but the simple fact of the matter remains that when you send skippy out to buy emergency jumpers or such, you’re most likely going to either get orange multimode or yellow singlemode and that’s just the way it is. Owen
On Mar 12, 2016, at 11:11 , Yardiel Fuentes <yardiel@gmail.com> wrote:
Hello Nanog-ers,
Have any of you had the option or; conversely, do you know of “best practices" or “common standards”, to color code physical cabling for your connections in DataCenters for Base-T and FX connections? If so, Could you share any ttype of color-coding schema you are aware of ?…. Yes, this is actually considering paying for customized color-coded cabling in a Data Center...
Mr. Google did not really provide me with relevant answers on the above… beyond the typical (Orange is for MMF, yellow for SMF, etc)…
Any reasons for or against it welcome too...
-- Yardiel Fuentes
Hi, What is the best solution for thin 2 mm or 0.9 mm fiber labelling? I like the idea of wrap around labels but does that work on thin wires? Maybe use something to pad the wire to more thickness where the label is to be? Regards, Baldur
Hi Baldur, Equinix in Sydney use the below, for Cross Connects. Goes around the fiber to pad it out, and the label keeps it on the fiber. http://www.cableorganizer.com/panduit/labelcore-cable-id-sleeve/ Been meaning to order some for internal use, too. Nick
The only problem I’ve had with those is that they tend to slide down the fiber and you can end up having to trace the fiber to find the label which sort of defeats the purpose. Owen
On Mar 13, 2016, at 18:33 , Nick Pratley <nick.pratley@serversaustralia.com.au> wrote:
Hi Baldur,
Equinix in Sydney use the below, for Cross Connects.
Goes around the fiber to pad it out, and the label keeps it on the fiber.
http://www.cableorganizer.com/panduit/labelcore-cable-id-sleeve/
Been meaning to order some for internal use, too.
Nick
Just place a piece of tape under the padding and it won't slide anymore. 5 seconds of extra work per end, though. On Sun, Mar 13, 2016 at 9:57 PM, Owen DeLong <owen@delong.com> wrote:
The only problem I’ve had with those is that they tend to slide down the fiber and you can end up having to trace the fiber to find the label which sort of defeats the purpose.
Owen
On Mar 13, 2016, at 18:33 , Nick Pratley < nick.pratley@serversaustralia.com.au> wrote:
Hi Baldur,
Equinix in Sydney use the below, for Cross Connects.
Goes around the fiber to pad it out, and the label keeps it on the fiber.
http://www.cableorganizer.com/panduit/labelcore-cable-id-sleeve/
Been meaning to order some for internal use, too.
Nick
-- :o@>
On Sun, 13 Mar 2016 22:21:48 -0400, "Oliver O'Boyle" said:
Just place a piece of tape under the padding and it won't slide anymore. 5 seconds of extra work per end, though.
I dunno. Your dexterity must be better than mine. I'd have trouble digging up the roll of tape, removing a section, putting the tape roll down, and applying the tape to the cable, all in 5 seconds. Especially if you drop it and it manages to bounce through a cutout in the raised floor. That's got to be the single best reason for overhead cabling. :)
Lol! I am very dextrous... But I prep by pulling off many pieces of tape at once and lining them up in advance. They don't need to go on perfectly. In fact, a few wrinkles help to keep the padding in place better than no wrinkles. Put a wire around the roll of tape and connect it to a small carabiner that you can clip to the rack or to other stable items wherever you're working (not individual cables in case you dislodge them). On Sun, Mar 13, 2016 at 11:58 PM, <Valdis.Kletnieks@vt.edu> wrote:
On Sun, 13 Mar 2016 22:21:48 -0400, "Oliver O'Boyle" said:
Just place a piece of tape under the padding and it won't slide anymore. 5 seconds of extra work per end, though.
I dunno. Your dexterity must be better than mine. I'd have trouble digging up the roll of tape, removing a section, putting the tape roll down, and applying the tape to the cable, all in 5 seconds.
Especially if you drop it and it manages to bounce through a cutout in the raised floor. That's got to be the single best reason for overhead cabling. :)
-- :o@>
On Mar 13, 2016, at 20:58 , Valdis.Kletnieks@vt.edu wrote:
On Sun, 13 Mar 2016 22:21:48 -0400, "Oliver O'Boyle" said:
Just place a piece of tape under the padding and it won't slide anymore. 5 seconds of extra work per end, though.
I dunno. Your dexterity must be better than mine. I'd have trouble digging up the roll of tape, removing a section, putting the tape roll down, and applying the tape to the cable, all in 5 seconds.
Especially if you drop it and it manages to bounce through a cutout in the raised floor. That's got to be the single best reason for overhead cabling. :)
Because it’s faster if you have to climb down off a ladder first before pulling up the floor tiles to track down the roll of tape? Oh, you mean that’s the single best reason for overhead COOLING. :P Owen
On Mon, 14 Mar 2016 11:15:29 -0700, Owen DeLong said:
On Mar 13, 2016, at 20:58 , Valdis.Kletnieks@vt.edu wrote: Especially if you drop it and it manages to bounce through a cutout in the raised floor. That's got to be the single best reason for overhead cabling. :)
Because it’s faster if you have to climb down off a ladder first before pulling up the floor tiles to track down the roll of tape?
Oh, you mean that’s the single best reason for overhead COOLING. :P
The tiles intended for cold air flow have lots of little 1/4" or so holes in them that a roll of tape can't fall through. The roll of tape *can* escape through the 6x6 cable cutout under the rack if there aren't too many cables in the way. If we had overhead cabling and under-floor cooling, we'd have no large cutouts anyplace. :) (Alas, the data center across the hall is 27 years old, and we'll need to build a new one to fix it. Someday soon, maybe. :)
On Mar 13, 2016, at 18:14 , Baldur Norddahl <baldur.norddahl@gmail.com> wrote:
Hi,
What is the best solution for thin 2 mm or 0.9 mm fiber labelling? I like the idea of wrap around labels but does that work on thin wires? Maybe use something to pad the wire to more thickness where the label is to be?
I doubt it would work on 0.9mm single strands, but I’ve had reasonably good luck with it on standard SM and MM fiber pairs with normal zip-cord style sheathing. Owen
Sun, Mar 13, 2016 at 05:10:26PM -0700, Owen DeLong wrote:
Whatever you do, please do not use Flag labels on cables… I HATE THEM. They are a constant source of entanglement and snags. They often get knocked off as a result or mangled beyond recognition, rendering them useless.
Hadn't seen that for ages, using Brother P-touch printers for small amounts of work and Avery Zweckform paper + laser printing for large cable installations. This stuff (Avery paper), http://computing.kiae.ru/~rea/wiring-porn-2.jpeg lives already for 5+ years and outlived many replacements of spine modules (that's InfiniBand fabric) and other operations without labels being ruined in any way. -- Eygene Ryabinkin, National Research Centre "Kurchatov Institute" Always code as if the guy who ends up maintaining your code will be a violent psychopath who knows where you live.
On Mar 14, 2016, at 04:42 , Eygene Ryabinkin <rea+nanog@grid.kiae.ru> wrote:
Sun, Mar 13, 2016 at 05:10:26PM -0700, Owen DeLong wrote:
Whatever you do, please do not use Flag labels on cables… I HATE THEM. They are a constant source of entanglement and snags. They often get knocked off as a result or mangled beyond recognition, rendering them useless.
Hadn't seen that for ages, using Brother P-touch printers for small amounts of work and Avery Zweckform paper + laser printing for large cable installations. This stuff (Avery paper), http://computing.kiae.ru/~rea/wiring-porn-2.jpeg lives already for 5+ years and outlived many replacements of spine modules (that's InfiniBand fabric) and other operations without labels being ruined in any way.
Sorry… To be clear, those hideous attrocities of P-Touch nastiness are exactly what I meant by flag labels. I had forgotten about the plastic zip ties with the over-sized surface on the ratchet. Compare to: http://www.cableorganizer.com/images/brady/brady-idxpert/labels-markers/imag... http://neumannmarking.com/wp-content/uploads/2015/03/135866255-300x199.jpg http://www.hotblog.co.uk/boblittlepr/wp-content/uploads/2013/06/cable-labels... Admittedly, I’m not wild about the bar-code scheme in the last one as I prefer human-readable labels, but YMMV. Owen
Just to throw it out there but I always try not to use RED cable. Normally, RED wire in any building is dedicated as FIRE system cabling. Curtis Starnes Senior Network Administrator Granbury ISD 600 W. Bridge St. Ste. 40 Granbury, Texas 76048 (817) 408-4104 (817) 408-4126 Fax curtis.starnes@granburyisd.org www.granburyisd.org OPEN RECORDS NOTICE: This email and responses may be subject to Texas Open Records laws and may be disclosed to the public upon request. -----Original Message----- From: NANOG [mailto:nanog-bounces@nanog.org] On Behalf Of Owen DeLong Sent: Sunday, March 13, 2016 7:10 PM To: Yardiel Fuentes <yardiel@gmail.com> Cc: nanog@nanog.org Subject: Re: DataCenter color-coding cabling schema I don’t know of any universal standards, but I’ve used the following in several installatins I was responsible for to good avail: Twisted Pair: RED: Untrusted Network (Internet or possibly DMZ) YELLOW: Optional for DMZ networks though I preferred to avoid documented in [1] below BLUE: Trusted Network (back-end, internal, etc.) GREEN: RS-232 straight-thru PURPLE: RS-232 X-Over (effectively Null Modem) 12345678 <-> 87654321 pin map. ORANGE: Ethernet X-Over (Best avoided documented in [2] below) GREY: Special purpose cabling not in one of the above categories Fiber: Orange — Multimode Fiber Yellow — Singlemode Fiber The absolute most useful thing you can do if you can impose the discipline to update the cable map rigorously and/or allocate manpower for periodic audits is to apply a unique serial number to each cable. I preferred to document not only the cable ID, but also the length. For the installations where I have worked, 5 digits was sufficient unique ID, so I used formats like IIIII-L[.L] where IIIII was a unique ID and L.L was the length of the cable in feet. (e.g. 00123-6.5 is cable number 123 which is 6.5 feet in length). The labels are (ideally) the self-laminating wrap-around types. I prefer the Brady labeling system which will automatically print 2-4 (depending on font size) instances of the label text on the self-laminating label such that it can be read from virtually any side of the cable without requiring you to rotate the label into view in most cases. The Brady labeling system is a bit overpriced compared to the Brother P-Touch, but the expanded capabilities and the quality of the label adhesives and such is, IMHO, sufficiently superior to justify the cost. Whatever you do, please do not use Flag labels on cables… I HATE THEM. They are a constant source of entanglement and snags. They often get knocked off as a result or mangled beyond recognition, rendering them useless. Similarly, I’ve found that circuit-ID and end-point labels on cables are often ill-maintained, so if you do use them, please make sure you remove them when the cable is moved/removed. The length is very useful because it gives you a radius within which the other end of the cable must be located and you can usually expect it to be reasonably close to the outer edge of that radius. More than a few times I’ve prevented a serious outage by giving the port number to the remote hands guy and then insisting that he read me the cable ID. “No, try the other port FE-0/2/4… You’re off by one. It’s above/left/right/below you.” [1] I prefer to avoid Yellow cables because some people have trouble understanding that Yellow Fiber and Yellow UTP might have different meanings. I also feel that the distinction between UNTRUSTED and DMZ networks is usually not all that important in most cabling situations. YMMV. [2] In this era of Auto-MDI/MDI-X ports and the like, it’s very rare to encounter a situation that truly requires a crossover cable with no viable alternative. If such is needed, I prefer to document it on the cable tags rather than using a special color code. Again, you have the risk of people not understanding that orange Fiber might not mean what Orange copper means. YMMV Yes, I know you can now get virtually any type of fiber in virtually any color, but the simple fact of the matter remains that when you send skippy out to buy emergency jumpers or such, you’re most likely going to either get orange multimode or yellow singlemode and that’s just the way it is. Owen
On Mar 12, 2016, at 11:11 , Yardiel Fuentes <yardiel@gmail.com> wrote:
Hello Nanog-ers,
Have any of you had the option or; conversely, do you know of “best practices" or “common standards”, to color code physical cabling for your connections in DataCenters for Base-T and FX connections? If so, Could you share any ttype of color-coding schema you are aware of ?…. Yes, this is actually considering paying for customized color-coded cabling in a Data Center...
Mr. Google did not really provide me with relevant answers on the above… beyond the typical (Orange is for MMF, yellow for SMF, etc)…
Any reasons for or against it welcome too...
-- Yardiel Fuentes
That's a good reason to use it. Who would cut it? ;) -A On Mon, Mar 21, 2016 at 8:53 AM, STARNES, CURTIS < Curtis.Starnes@granburyisd.org> wrote:
Just to throw it out there but I always try not to use RED cable. Normally, RED wire in any building is dedicated as FIRE system cabling.
Curtis Starnes Senior Network Administrator Granbury ISD 600 W. Bridge St. Ste. 40 Granbury, Texas 76048 (817) 408-4104 (817) 408-4126 Fax curtis.starnes@granburyisd.org www.granburyisd.org
OPEN RECORDS NOTICE: This email and responses may be subject to Texas Open Records laws and may be disclosed to the public upon request.
-----Original Message----- From: NANOG [mailto:nanog-bounces@nanog.org] On Behalf Of Owen DeLong Sent: Sunday, March 13, 2016 7:10 PM To: Yardiel Fuentes <yardiel@gmail.com> Cc: nanog@nanog.org Subject: Re: DataCenter color-coding cabling schema
I don’t know of any universal standards, but I’ve used the following in several installatins I was responsible for to good avail:
Twisted Pair:
RED: Untrusted Network (Internet or possibly DMZ) YELLOW: Optional for DMZ networks though I preferred to avoid documented in [1] below BLUE: Trusted Network (back-end, internal, etc.) GREEN: RS-232 straight-thru PURPLE: RS-232 X-Over (effectively Null Modem) 12345678 <-> 87654321 pin map. ORANGE: Ethernet X-Over (Best avoided documented in [2] below) GREY: Special purpose cabling not in one of the above categories
Fiber: Orange — Multimode Fiber Yellow — Singlemode Fiber
The absolute most useful thing you can do if you can impose the discipline to update the cable map rigorously and/or allocate manpower for periodic audits is to apply a unique serial number to each cable. I preferred to document not only the cable ID, but also the length. For the installations where I have worked, 5 digits was sufficient unique ID, so I used formats like IIIII-L[.L] where IIIII was a unique ID and L.L was the length of the cable in feet. (e.g. 00123-6.5 is cable number 123 which is 6.5 feet in length).
The labels are (ideally) the self-laminating wrap-around types. I prefer the Brady labeling system which will automatically print 2-4 (depending on font size) instances of the label text on the self-laminating label such that it can be read from virtually any side of the cable without requiring you to rotate the label into view in most cases.
The Brady labeling system is a bit overpriced compared to the Brother P-Touch, but the expanded capabilities and the quality of the label adhesives and such is, IMHO, sufficiently superior to justify the cost.
Whatever you do, please do not use Flag labels on cables… I HATE THEM. They are a constant source of entanglement and snags. They often get knocked off as a result or mangled beyond recognition, rendering them useless.
Similarly, I’ve found that circuit-ID and end-point labels on cables are often ill-maintained, so if you do use them, please make sure you remove them when the cable is moved/removed.
The length is very useful because it gives you a radius within which the other end of the cable must be located and you can usually expect it to be reasonably close to the outer edge of that radius.
More than a few times I’ve prevented a serious outage by giving the port number to the remote hands guy and then insisting that he read me the cable ID. “No, try the other port FE-0/2/4… You’re off by one. It’s above/left/right/below you.”
[1] I prefer to avoid Yellow cables because some people have trouble understanding that Yellow Fiber and Yellow UTP might have different meanings. I also feel that the distinction between UNTRUSTED and DMZ networks is usually not all that important in most cabling situations. YMMV.
[2] In this era of Auto-MDI/MDI-X ports and the like, it’s very rare to encounter a situation that truly requires a crossover cable with no viable alternative. If such is needed, I prefer to document it on the cable tags rather than using a special color code. Again, you have the risk of people not understanding that orange Fiber might not mean what Orange copper means. YMMV
Yes, I know you can now get virtually any type of fiber in virtually any color, but the simple fact of the matter remains that when you send skippy out to buy emergency jumpers or such, you’re most likely going to either get orange multimode or yellow singlemode and that’s just the way it is.
Owen
On Mar 12, 2016, at 11:11 , Yardiel Fuentes <yardiel@gmail.com> wrote:
Hello Nanog-ers,
Have any of you had the option or; conversely, do you know of “best practices" or “common standards”, to color code physical cabling for your connections in DataCenters for Base-T and FX connections? If so, Could you share any ttype of color-coding schema you are aware of ?…. Yes, this is actually considering paying for customized color-coded cabling in a Data Center...
Mr. Google did not really provide me with relevant answers on the above… beyond the typical (Orange is for MMF, yellow for SMF, etc)…
Any reasons for or against it welcome too...
-- Yardiel Fuentes
Good point, never looked at it that way, but I have had techs before that would cut anything they thought was data and sometimes even when they knew it was not. I guess it was Beer:30 time to them :-\ Curtis From: Aaron C. de Bruyn [mailto:aaron@heyaaron.com] Sent: Monday, March 21, 2016 2:45 PM To: STARNES, CURTIS <Curtis.Starnes@granburyisd.org> Cc: Owen DeLong <owen@delong.com>; Yardiel Fuentes <yardiel@gmail.com>; nanog@nanog.org Subject: Re: DataCenter color-coding cabling schema That's a good reason to use it. Who would cut it? ;) -A On Mon, Mar 21, 2016 at 8:53 AM, STARNES, CURTIS <Curtis.Starnes@granburyisd.org<mailto:Curtis.Starnes@granburyisd.org>> wrote: Just to throw it out there but I always try not to use RED cable. Normally, RED wire in any building is dedicated as FIRE system cabling. Curtis Starnes Senior Network Administrator Granbury ISD 600 W. Bridge St. Ste. 40 Granbury, Texas 76048 (817) 408-4104<tel:%28817%29%20408-4104> (817) 408-4126<tel:%28817%29%20408-4126> Fax curtis.starnes@granburyisd.org<mailto:curtis.starnes@granburyisd.org> www.granburyisd.org<http://www.granburyisd.org> OPEN RECORDS NOTICE: This email and responses may be subject to Texas Open Records laws and may be disclosed to the public upon request. -----Original Message----- From: NANOG [mailto:nanog-bounces@nanog.org<mailto:nanog-bounces@nanog.org>] On Behalf Of Owen DeLong Sent: Sunday, March 13, 2016 7:10 PM To: Yardiel Fuentes <yardiel@gmail.com<mailto:yardiel@gmail.com>> Cc: nanog@nanog.org<mailto:nanog@nanog.org> Subject: Re: DataCenter color-coding cabling schema I don’t know of any universal standards, but I’ve used the following in several installatins I was responsible for to good avail: Twisted Pair: RED: Untrusted Network (Internet or possibly DMZ) YELLOW: Optional for DMZ networks though I preferred to avoid documented in [1] below BLUE: Trusted Network (back-end, internal, etc.) GREEN: RS-232 straight-thru PURPLE: RS-232 X-Over (effectively Null Modem) 12345678 <-> 87654321 pin map. ORANGE: Ethernet X-Over (Best avoided documented in [2] below) GREY: Special purpose cabling not in one of the above categories Fiber: Orange — Multimode Fiber Yellow — Singlemode Fiber The absolute most useful thing you can do if you can impose the discipline to update the cable map rigorously and/or allocate manpower for periodic audits is to apply a unique serial number to each cable. I preferred to document not only the cable ID, but also the length. For the installations where I have worked, 5 digits was sufficient unique ID, so I used formats like IIIII-L[.L] where IIIII was a unique ID and L.L was the length of the cable in feet. (e.g. 00123-6.5 is cable number 123 which is 6.5 feet in length). The labels are (ideally) the self-laminating wrap-around types. I prefer the Brady labeling system which will automatically print 2-4 (depending on font size) instances of the label text on the self-laminating label such that it can be read from virtually any side of the cable without requiring you to rotate the label into view in most cases. The Brady labeling system is a bit overpriced compared to the Brother P-Touch, but the expanded capabilities and the quality of the label adhesives and such is, IMHO, sufficiently superior to justify the cost. Whatever you do, please do not use Flag labels on cables… I HATE THEM. They are a constant source of entanglement and snags. They often get knocked off as a result or mangled beyond recognition, rendering them useless. Similarly, I’ve found that circuit-ID and end-point labels on cables are often ill-maintained, so if you do use them, please make sure you remove them when the cable is moved/removed. The length is very useful because it gives you a radius within which the other end of the cable must be located and you can usually expect it to be reasonably close to the outer edge of that radius. More than a few times I’ve prevented a serious outage by giving the port number to the remote hands guy and then insisting that he read me the cable ID. “No, try the other port FE-0/2/4… You’re off by one. It’s above/left/right/below you.” [1] I prefer to avoid Yellow cables because some people have trouble understanding that Yellow Fiber and Yellow UTP might have different meanings. I also feel that the distinction between UNTRUSTED and DMZ networks is usually not all that important in most cabling situations. YMMV. [2] In this era of Auto-MDI/MDI-X ports and the like, it’s very rare to encounter a situation that truly requires a crossover cable with no viable alternative. If such is needed, I prefer to document it on the cable tags rather than using a special color code. Again, you have the risk of people not understanding that orange Fiber might not mean what Orange copper means. YMMV Yes, I know you can now get virtually any type of fiber in virtually any color, but the simple fact of the matter remains that when you send skippy out to buy emergency jumpers or such, you’re most likely going to either get orange multimode or yellow singlemode and that’s just the way it is. Owen
On Mar 12, 2016, at 11:11 , Yardiel Fuentes <yardiel@gmail.com<mailto:yardiel@gmail.com>> wrote:
Hello Nanog-ers,
Have any of you had the option or; conversely, do you know of “best practices" or “common standards”, to color code physical cabling for your connections in DataCenters for Base-T and FX connections? If so, Could you share any ttype of color-coding schema you are aware of ?…. Yes, this is actually considering paying for customized color-coded cabling in a Data Center...
Mr. Google did not really provide me with relevant answers on the above… beyond the typical (Orange is for MMF, yellow for SMF, etc)…
Any reasons for or against it welcome too...
-- Yardiel Fuentes
On Sat, Mar 12, 2016 at 2:11 PM, Yardiel Fuentes <yardiel@gmail.com> wrote:
Have any of you had the option or; conversely, do you know of “best practices" or “common standards”, to color code physical cabling for your connections in DataCenters for Base-T and FX connections? If so, Could you share any ttype of color-coding schema you are aware of ?…. Yes, this is actually considering paying for customized color-coded cabling in a Data Center...
Mr. Google did not really provide me with relevant answers on the above… beyond the typical (Orange is for MMF, yellow for SMF, etc)…
Any reasons for or against it welcome too...
Hi Yardiel, Patch cables or fixed cabling to patch panels? For fixed cabling, it's common to pick colors which match the cable type. Orange for multimode fiber, yellow for single mode fiber, blue for four-pair cat5e, something else for cat6, etc. At each end, label the location of the opposite endpoint twice, once on the panel and once on the cable itself (cables can pull loose from panels). With fixed cabling terminating in patch panels they'll tend to get reused over time for different types of signalling so don't overthink it. For patch cables, it's common to pick a color for each type of physical signaling so you don't jam the wrong kind of signal in to a port that doesn't match. Your gig-e switch may not like the voltage from that ringing pots line. Blue for ethernet, white for POTs, green for T1s, some other color for the rs232 serial cables, IP/KVM cables, etc. I find the easiest way to label patch cables is with color electrical tape. Put the same bands of unique colors at both ends of the cable. This will let you visually identify the cables without pulling on them to try and line up tiny text on the tags with your eyeballs. Regards, Bill Herrin -- William Herrin ................ herrin@dirtside.com bill@herrin.us Owner, Dirtside Systems ......... Web: <http://www.dirtside.com/>
On 14 March 2016 at 00:23, William Herrin <bill@herrin.us> wrote:
On Sat, Mar 12, 2016 at 2:11 PM, Yardiel Fuentes <yardiel@gmail.com> wrote:
Have any of you had the option or; conversely, do you know of “best practices" or “common standards”, to color code physical cabling for your connections in DataCenters for Base-T and FX connections?
For patch cables, it's common to pick a color for each type of physical signaling
I used to support this view too, but over the last few years, as everything has (basically) become Ethernet, I've taken to a different scheme. For copper patching, I now recommend my clients simply invest in a range of colored patch cables and use them randomly. The length of the patch cable is much more important than the color (too little length will make it difficult to re-route cables if you need to remove cards etc. and too long will mean tangles and space taken up with loops of excess cable.) The benefits of my "rainbow" scheme are: 1. easier to identify both ends of a cable, reducing disconnect errors. When tracing a cable in a bundle or on a patch bay, it's easy when they're different colors. 2. no need to police the cable scheme - if you have a strict color regime, what do you do when someone uses the wrong color? especially if a disconnect would be service affecting. It's really hard to justify "maintenance downtime" to an account manager on the basis of you not liking the color of a patch cable. Aled
Hi, I'm not sure I'm keen on a colour standard - especially given our recent difficulties sourcing cabling to our spec in certain colours...or lengths! however, what we do - and others do based on this thread - is have our own internal colour scheme for purposes/systems/customers. fibre is far more difficult for this - coloured labels (and a decent labelling regime in the first place) win in that arena. (obviously the copper plant has labelling too but the choice of colours means that function/purpose is already known from many metres away ;-) ) alan
On Mar 14, 2016, at 03:15 , Aled Morris <aledm@qix.co.uk> wrote:
On 14 March 2016 at 00:23, William Herrin <bill@herrin.us> wrote:
On Sat, Mar 12, 2016 at 2:11 PM, Yardiel Fuentes <yardiel@gmail.com> wrote:
Have any of you had the option or; conversely, do you know of “best practices" or “common standards”, to color code physical cabling for your connections in DataCenters for Base-T and FX connections?
For patch cables, it's common to pick a color for each type of physical signaling
I used to support this view too, but over the last few years, as everything has (basically) become Ethernet, I've taken to a different scheme.
For copper patching, I now recommend my clients simply invest in a range of colored patch cables and use them randomly.
The length of the patch cable is much more important than the color (too little length will make it difficult to re-route cables if you need to remove cards etc. and too long will mean tangles and space taken up with loops of excess cable.)
The benefits of my "rainbow" scheme are:
1. easier to identify both ends of a cable, reducing disconnect errors. When tracing a cable in a bundle or on a patch bay, it's easy when they're different colors.
But if you serialize them and have the company you order the cables from do the labeling (I’ve had this done, it’s not difficult and doesn’t add significantly to the cost of the cables), then that’s even more useful for that purpose than your “rainbow” scheme.
2. no need to police the cable scheme - if you have a strict color regime, what do you do when someone uses the wrong color? especially if a disconnect would be service affecting. It's really hard to justify "maintenance downtime" to an account manager on the basis of you not liking the color of a patch cable.
Well… 1. You have someone whose responsibility it is to keep appropriate sized cables of various colors in stock so that there’s no incentive to do so. 2. You don’t let untrained monkeys play in your cage. 3. You hand the 1d10t that did it a roll of appropriately colored electrical tape and provide instructions on how to spiral-wrap. He gets to change the color of the cable. (This will make a repeat offense relatively unlikely as it’s a huge PITA). Owen
----- Original Message -----
From: "Yardiel Fuentes" <yardiel@gmail.com>
Have any of you had the option or; conversely, do you know of “best practices" or “common standards”, to color code physical cabling for your connections in DataCenters for Base-T and FX connections? If so, Could you share any ttype of color-coding schema you are aware of ?…. Yes, this is actually considering paying for customized color-coded cabling in a Data Center...
EIA/TIA 568/598 talk to fiber color coding. They have one for copper too, but I can't find it either, off hand. google://cable+jacket+color+standards is a pretty good search for this Cheers, -- jra -- Jay R. Ashworth Baylink jra@baylink.com Designer The Things I Think RFC 2100 Ashworth & Associates http://www.bcp38.info 2000 Land Rover DII St Petersburg FL USA BCP38: Ask For It By Name! +1 727 647 1274
participants (15)
-
A.L.M.Buxey＠lboro.ac.uk -
Aaron C. de Bruyn -
Aled Morris -
Baldur Norddahl -
Eygene Ryabinkin -
Jay Hennigan -
Jay R. Ashworth -
Joe Hamelin -
Nick Pratley -
Oliver O'Boyle -
Owen DeLong -
STARNES, CURTIS -
Valdis.Kletnieks＠vt.edu -
William Herrin -
Yardiel Fuentes