Ask the Expert: Customers Save Time with Optic Cable Cleaning

If there is a lot of cables to clean, may take a long time, in your experience, how much time on average does it take to clean a cable, both end?

ka_man, good question, it really only takes  a minute or two max per end face.  If that is done in a diagnostic situation then you are normally only looking a few ends.  If it is as part of an install then the added time compared to the time you can save further down the line in minimal.    Take a look at Bill's video demo that will give you a good idea.

https://youtu.be/u1rgjzUQlDU

Hi Roberto,

I wanted to respond to your question about the real benefits of optic cable cleaning.

Optic cable cleaning facilitates customer education and self service.  Customers requesting a replacement may wait over a day for an engineer to be scheduled and to appear on site.  Optic cable cleaning could remediate the issue in a couple of minutes.

John

well, unplug , clean, unplug clean again will be way too disruptive to hosts. So i would rather wait for another SFP, clean patches an go from there.

Thanks for the reply.  We are talking about cleaning as a step in the troubleshooting process.  Our testing has found that 85% of the SFPs returned were no fault found.  With the Connectrix switches, you can disconnect the cable where an error (yellow light) is displayed, clean the cable and possibly restore connectivity without have to replace the SFP.

Hi experts, we've talked about the very short time it takes (< 2min) to clean an optic cable.  Since we find optic cables in any device (server, storage, network) that connects into the network, who could perform the Optic Cable Cleaning procedure?

Indeed and to replace the SFP the cable needs to be removed anyway so as far as service interruption goes it not that much more and statistically cleaning is far more likely to resolve the issue that swapping an SFP out.  The stats speak fpr themselves, 80% chance the end face is contaminated, either root cause or contributory, or 5%-10% chance the SFP itself is root cause. It is just we have all become conditioned to call out SFP swap as in initial diagnostic step.  So it has become ingrained in the data center playbook.

it is extremely rear that we have problems with fiber cables themselves. Most of the time it's a SFP on VNX/VMAX going bad and typically it's hard down, not bouncing up/down or generating tons of CRC errors on switch ports. There have been one or two occasions where we had ports on MDS go down with "Port error threshold exceeded". I look at the port and see a bunch of CRC errors so i go ahead and swap SFP first and clear port counters.  If counters start going up again I go ahead and replace everything in the data path, I don't want to keep bouncing port for hosts (today let's replace this patch, hmm not it, let's try this patch now ..hmm no it).  See what i mean ?

Hey  Brian, you and I must have the same CE.  Like a bull in a china shop, ripping through my aqua spaghetti, trying to replace a module on VNX  hahaha

Dynamox, are you having this issue?  90% of my F/C problems were at the patch panel OR someone bent or stepped on the F/C cable.   I'm sure you traced the cable(s) right?

Haha amen!  How many curves/bends are in the cables matter too.  Most people who lay F/C cable tend to make short/tight bends. That's also a no-no!   Easy-longer curves are better.

You running long cables or short cables?  I'm actually curious now as to what your setup is over there?

Please chime in....this is a great discussion.

Jford wrote: Thanks for the reply.  We are talking about cleaning as a step in the troubleshooting process.  Our testing has found that 85% of the SFPs returned were no fault found.  With the Connectrix switches, you can disconnect the cable where an error (yellow light) is displayed, clean the cable and possibly restore connectivity without have to replace the SFP.

for a single host, sure I might do it. For an SFP that connects to a UCS enclosure or a VMAX FA i am not taking multiple disruptions to my hosts to "experiment".  Is it the SFP, is it the cable, is it the patch panel.  Hammer approach ? Sure but I can't keep bouncing connectivity for my hosts while i am playing whack a mole

For sure.  Its even rare for me to see SFPs going bad but they do.  You are correct on the CRC errors as well. I guess what I was going for here is for example, I have had issues with CE's going on to replace hardware only to bump/step/mash a F/C cable.  Happened a few months to me ("dude! watch what you're doing!") so that is where I was headed.  If you changed a SFP but are still getting errors, I would trace that cable. All it takes is a nick in the cable nick or what not.

Sorry for chiming in, I know I this can be frustrating!