Minimum cable distance needed for CRC?

[LynK]

Is there such thing as a minimum distance on a cable, in which is needed to perform a CRC check?

is it different between cable/fiber? I am not aware of any limitations for CRC

[Otanx]

So shooting from the hip. I would think for full duplex copper there is no minimum distance. Half duplex there would be a minimum distance based on how long a NIC can transition from sending data to listening for the collision. Optical cables the minimum distance is dependent on the optics. The transmitter in some optics have a power level high enough to damage a receiver, and attenuation over distance brings the level down so the receiver can see it. Even before you get to power levels that will damage the receiver you can still have a power level high enough to blind the receiver without damage. There is a NANOG presentation that covers all the optical stuff somewhere. It is a very interesting read.

-Otanx

[deanwebb]

I've heard 3' as the minimum length on full duplex copper.

[killabee]

Based on this, not for 10BASE-T, 100BASE-T, and 1000BASE-T.

For fiber, what Otanx said....though I still wonder if there are optics out there that will tell the sender that its Tx power is too strong and to lower its...that's a question for javentre

[deanwebb]

Based on this, not for 10BASE-T, 100BASE-T, and 1000BASE-T.

For fiber, what Otanx said....though I still wonder if there are optics out there that will tell the sender that its Tx power is too strong and to lower its...that's a question for javentre

LOL, it was an old guy that said it.

[javentre]

I had this discussion with a number of folks back in 2008, regarding copper cables.  I thought it was 1', other folks agreed, but no one could point to something as proof.  I was unable to find any standards based documentation which stipulated a minimum cable length.

With optical links, if you have high power optics (ER/ZR) with short spans, you often need to use pads to attenuate the signal so you don't burn out the receivers.   Most of the Cisco 10G-SR transceivers have a maximum TX that is weaker than the maximum RX, so you'll never need to attenuate them, assuming the optics are in spec.  For example, most Cisco 10G SR optics have a maximum TX of -1.2 but the maximum Rx is -1.0, so you're good.  For 10G LR, the max TX and RX are the same, so you're fine most of the time.

.I still wonder if there are optics out there that will tell the sender that its Tx power is too strong and to lower its...that's a question for javentre

You'll want to read about active DWDM systems, specifically something called APC:)
http://www.cisco.com/c/en/us/td/docs/optical/15000r9_0/dwdm/reference/guide/454d90_ref/454d90_networkref.html#wp335800

[LynK]

What I am most concerned about is 2 ASAs running 1G with GLC-SX-MM transceivers with 62.5/125 fiber. The whole reason I am curious is we have 2 ASA 5525x's which are using gi1/5 (1G fiber) for the failover interface between the two. Right now they are literally 2u apart from one another. Currently we have a 3m cable interconnecting the two, but if I could get a .3m cable and there would be no issues with CRC/failover i would prefer to go the route with no slack.

[javentre]

You're fine.

1000SX is even better than 10G.   TX maximum is -3 and RX maximum is 0.

[Otanx]

NANOG fiber presentation - https://www.nanog.org/meetings/nanog48/presentations/Sunday/RAS_opticalnet_N48.pdf

-Otanx