cat6 not working w/ T1 CSU/DSU on 2600??

[Cooky]

We're seeing a lot of errors on the serial interface.
Carrier points finger at our cable, citing they were able to loop lean to the smartjack.

We replaced cable between T1 CSU/DSU and smartjack w/ a brand new cat 6 cable.
Now carrier is saying it became even worse, and wants us to try another cable.

WTF??
Is this possible? Cat 6 not compatible w/ T1?!
Can't they consider the posibility of the CSU/DSU going bad?
They claimed they already reset the interface, and diagnostic came back clean...
So now we're pointing finger at each other.

Anyone know how I can win this battle?

 

[p0lar]

Originally posted by: Cooky
We're seeing a lot of errors on the serial interface.
Carrier points finger at our cable, citing they were able to loop lean to the smartjack.

We replaced cable between T1 CSU/DSU and smartjack w/ a brand new cat 6 cable.
Now carrier is saying it became even worse, and wants us to try another cable.

WTF??
Is this possible? Cat 6 not compatible w/ T1?!
Can't they consider the posibility of the CSU/DSU going bad?
They claimed they already reset the interface, and diagnostic came back clean...
So now we're pointing finger at each other.

Anyone know how I can win this battle?

How long is the total loop? Are you using shielded cable?

If they're still arguing it's the cable, they're probably 99% correct. Seriously, I sound like a broken record, but most problems are cable or connector-related. Have them replace the cable for you, they'll stop pointing the finger quickly thereafter.

 

[Cooky]

All the equipment is located in a access restricted wiring closet w/ AC & backup power.
You're telling me there's more chance for a cable to go bad by itself than a T1 CSU/DSU card and/or a router?

If this was a new install, I agree w/ you that we need to look at every detail.
This circuit has been running error free until now.

 

[ScottMac]

Nope, UTP of any sort is not spec for T1.

Many of the CSU/DSU manufacturers will tolerate UTP ( Cat{anything} ) for shorter runs, but for anything over ~12 feet, it's a crap shoot.

What you are looking for used to be called "Premises Cable" (now just called "T1" cable, at least in the Anixter catalog (www.anixter.com).

Prem cable is individually shielded pair in an overall shielded sheath.

"Why won't Cat 5 / 5e / 6 work well @ 1.5 mbps when it does so well at 100Mbps?", I hear you ask?

It's the type of signaling. T1 is straight up baseband pulses ... they go up, or they go down, but they are single (very noisy, high crosstalk) pulses. Ethernet uses a differential signaling (Manchester encoding? ) and it reduces the sensitivity to crosstalk and common mode noise.

"But, they DELIVER the T1/DS1 on UTP to my prem, why does THAT work?" I think I also hear you asking .... well, the signaling up to the SJ/NIU is HDSL, not "T1." Part of the SJ/NIU's job is to convert the HDSL into DSX1 ("real DS1/T1 signaling") to be delivered to your CSU/DSU on Prem cabling.

As p0lar asked, how long is the cable between the SJ/NIU and the CSU/DSU?

If your Cat6 wasn't terminated properly (like a commercial / "store bought" cable), then it's all the more susceptable to crosstalk.

If you want to prove this out, move the router (and / or CSU/DSU) closer to the SJ/NIU and try a shorter (UTP) cable.

If you're interested in the hard spec, check out the Telecordia specification for DSX1/DS1/T1 ....

Good Luck

Scott

 

[spidey07]

ScottMac says it all.

You are out of spec.

 

[cross6]

Good info in this thread

 

[Cooky]

Thank you ScottMac for the info.
I knew we were supposed to use T1 cables, but not all the details and the reasons behind.
Unfortunately our remote sites usually don't have spare T1 cables laying around to convince carrier it's not the cable.

Were we out of spec? Yes.
Was the carrier full of crap and lying out of their teeth? Also a yes.

We use Vanco as our network opoerator because with the number of sites we have to support, it's not feasible to do everything in house.
Everytime we call them, they'd try to blame fault on the cables. This just pisses me off.
We've had outage in three different sites in the past two days. They blamed on the cables as usual for all three of them.

Network for the first site just came back up all of a sudden. Asked them for the RFO, all they can say is "error cleared while testing by carrier", one of the other usual crap I hate about them.

Second site was due to fault on one of the smartjacks.

They RMA'd the T1 WIC for the third site, and we'll know the outcome tomorrow morning.

Instead of chasing the carriers for us as they're supposed to, they just blame everything on us...a lot of times their SMARTS system won't even detect an outage. (or sometimes the other way around - send out alerts while the site has been ceased for months)
We'd call them AFTER users from the site report it.

There was a reason behind my frustration as you can see...

 

[jlazzaro]

i feel your pain cooky!

you say its not feasible to manage in house...just curious, how many sites do you support?

 

[Cooky]

~50

Edit-
We have 6 network engineers (me included) to support around 50 sites.

 

[ScottMac]

I understand your frustration, truely I do.

And, do not in any way take this as defense for a provider that is not "providing" ... but it really does happen on occasion, much more often that you'd think, that the line "comes clear while testing."

Sometimes some component in the network loses sync or gets confused. During testing, a variety of bit patterns are sent down the line, usually to / through a loop, and back to the originator to compare what was sent versus what was received (I know you know this, but for other folks that don't).

Occasionally, that stream of testing bits shakes the component back into sync, and the circuit comes back up.

Given that that is is case, you should keep an eye on that circuit, because, likely as not, it has a weak component (PLL, receiver, amp, equalizer, etc) that will probably drop out again in the future ... not always, but probably better'n fifty/fifty.

One of my customers has had a site down (on BRI backup) for nearly a week. Telco kept testing, we swapped the WIC, the (external) CSU/DSU, had the cabling checked, proved the "back of the SJ" (hard loop on the customer side back to Telco) ... NNI frame checks ... everything tested good .... perfect in fact ... but it didn't work. Earlier in the week, Telco *did* find a bad repeater ... but the system was still down.

Tonight, in a group call, we figured out that the problem was that the local frame provider forgot to turn LMI back on after a test. Everything tested 100% OK, because the test patterns don't use LMI ... but once everything was unlooped & normalized, the router wouldn't come up (no LMI).

In this case (Cisco router), DEBUG FRAME ADJ PVC and DEBUG FRAME LMI was our friend. Look for the "myseq ##" and "yourseq ##" parameters; they showed we were putting LMI out, but didn't get anything back from the cloud.

Sometimes the problem is basic, sometimes it's a real ass-cruncher ... that's the challenge. Standard rules apply ... start at layer one and work your way up.

Good Luck with your stuff

FWIW

Scott

 

[spidey07]

Originally posted by: ScottMac
In this case (Cisco router), DEBUG FRAME ADJ PVC and DEBUG FRAME LMI was our friend. Look for the "myseq ##" and "yourseq ##" parameters; they showed we were putting LMI out, but didn't get anything back from the cloud.

More awesome info from ScottMac. Although he's moved up into layer2.

I don't have much to add other than work the layers. Always start at L1, once you have confirmed beyond any doubt that is good then move up - L1 means cabling/framing/clocking/sync and it truly is the prime cause of most all network problems. "cleared while testing" is many times a cop-out. But it does truly happen to re-sync 'sometimes', other times it's a telco's way of saying "we fixed our stuff"

What I will say is outsourcing your WAN to somebody other than those who have DIRECT access into problem resolution is a bad idea all around.

From what cooky has posted he's got a "we give you an ethernet port and take care of the rest". That's a great way to do things, but with 6 network folks at only 50 sites he can take care of it. We have 3 guys for 600 sites.

As I've posted before, it's all about account management. I can go off on a small rant like I did in OT, but you get what you pay for. Labor/people is expensive and hits every year. Good design saves money.