BlueWolf's Howl

« Loopbacks revisited | Bluewolf's Howl | RSS Madness »

October 19, 2005

Telco translations

Quite often, I'll be involved in a 'vendor meet.' In network parlance, it means that a circuit is not working and everyone is shrugging their shoulders and saying, "it's not on my end, check with ___." Usually that means that there's trouble with the circuit, but the telco won't admit/believe it.

Granted, they probably have had their share of customers who call to have the circuit tested or report it down ... only to find that the router is unplugged or someone altered a configuration, but the meets I've been on are a result of the telco not believing the customer's complaint. They test out their end and it 'proves out good'...and the telco tests their end and it 'proves out good,' yet the line still isn't operational.

I've always suspected there was something more to it. No, not anything sinister. And before you get out the white jacket with the really long sleeves, no, I don't think it's any kind of anti-customer conspiracy, either. What I suspected was that there was a place between the network and the telco and it was like a crack in the mirror that was so thin that no one could see it, but everyone could feel it.

When I run into a telco guy, I always try to find out a little bit more about the mysterious world of telco. No, I haven't run into any female telco techs, and there are many other similarities with the Masons, but that's another post. When I first started asking questions, the first response I got was very similar to "don't strain your brain thinking about it, little lady." So I adjusted my questioning and the response changed to something like "I'm not going to answer your scab question and I'm calling my union steward on this line just to prove it works." I readjusted my approach, yet again, to accomodate the fragile male ego and job insecurity. I got one step closer.

The nice telco tech I last asked actually tried to answer my questions. I wanted to know what they're actually testing when they do their magical test. When they say "the line's good," what do they really mean? What layer are they testing and how? Unfortunately, he didn't really understand what I was asking. Layers 1, 2 and 3 of the OSI model don't really mean anything to a telco guy. But, I found out that he's testing the line not just by the voltage, but also by "sending patterns towards the loop." Yeah, I caught them many times before with their little voltage meters - looking at the voltage on the line. I still don't know what that was all about yet, but it looked offical and definitive.

So I decided to take a break from restudying the BCRAN material and do some leisure reading. Okay, so you might not think it's leisure reading, but the O'Reilly book: "T1 A Survival Guide" seemed like a nice break to me. Right there in Chapter 4 (page 25), I found out exactly what kind of tests they're performing.

When a T1 is installed, the telco technician may hook a handheld device with lots of buttons and blinking lights on it up to the new T1 jack. The testing device can perform stress tests on the new line to measure its quality and clarity. After looping back to the T1, the testing device injects a specific bit stream. Returning bits are compared against the original sequence to determine the bit error rate (BER).

Ah...so that's what they're doing. They're stressing the line in different ways to rule out/rule in various possible causes of a 'bad line.' And to answer my own initial question...they're testing the physical layer of the T1 line. Now, let's look deeper...

The QRS test (quasi random signal) is a test pattern in a sequence that is not guaranteed to meet pulse-density requirements. On B8ZS links, the CSU/DSU will perform zero substitution and the transmitted signal will meet pulse-density requirements without a problem. AMI-encoded links should not alter the signal to meet minimum pulse-density requirements, but misconfigured equipment on the line may be configured to do so. Errors observed in the QRS test indicate one of two things: a line that is bad, or a line that has misconfigured equipment. One common source of misconfiguration is related to pulse stuffing on the CSU/DSUs or other line components.

Wow. That's a lot of useful information! If this test failed and the telco guy finds that it's not a 'bad line,' but a 'misconfiguration,' at least I know where to look. I would need to look at the CSU/DSU and see if it's bit stuffing. On an Adtran TDU 120e, there is a menu option that can either enable or disable bit stuffing (Chapter 5 of the Device Manual). On an Adtran 600 or 600e, there is a bit stuffing configuration option under the Network menu (also in Chapter 5 of that manual). This is much more helpful than "I'm sorry, but it's somewhere in your equipment."

Other common tests include:
3-in-24 test for AMI-encoded links. This stresses the link by sending the lowest density signal allowed by the specifications.
1-in-8 test for B8ZS-encoded links. Same as above - stresses the link. If zeros are creeping into the pattern, it will perform a zero substitution when its not supposed to do so.
All-zeros test - for B8ZS which should result in constant substitution.
All-ones test - for both B8ZS and AMI links. This is the test result that you'll want to ask about.

This test checks for ringing and crosstalk. Ringing occurs when transmitted signals reflect off boundaries and the reflections bounce back and interfere with later transmitted signals. Crosstalk occurs when the cable pairs are not separated correctly, so that the transmitted signal induces a similar signal in the receive pair.

Then again...if the smart jack doesn't have any lights on it at all, the smart jack itself is bad. Yes, I've seen where the telco says that their equipment tested clean to the smart jack and the smart jack itself was completely dead. Wonder what kind of tests they do?

It reminds me of the tech that told me that his "telnet is dropping off" at a certain IP address. Hmmm. Did he digitize himself and follow the packets? Telnet either connects or doesn't. It doesn't drop off at a certain point. What he was actually doing was trying to telnet and then performing a tracert to supposedly figure out where his telnet connection was dying. ICMP was filtered at the firewall. Telnet wasn't. I was able to telnet out to the Internet and then telnet back in - telnet was working. It was a Layer 8 problem (problem exists between keyboard and chair).

So now I've solved (for myself) the mystery of the T1 pattern testing. If you're still curious about smart jacks and what those blinking lights mean... You can find more information about the LEDs from the Adtran site. You'll find the manuals (and menu options) for their brand of CSU/DSUs and smart jack cards. Those cards are actually called "TSU IQ Rackmount" devices. And the little locked box where they live is called a "Smart 16e Shelf" (which also has a manual on the site).

Now that I speak a little Telco, maybe I'm less likely to be the Layer 8 problem at the demarc...

Posted by BlueWolf on October 19, 2005 07:47 PM