jhh@ihlpl.UUCP (Haller) (10/23/87)
The last time someone fought the battle of the "}", it was discovered that there was a synchronization problem in digital facilities somewhere between the two end points. The 212 modem is particularly sensitive to phase shifts, causing bit errors. Normal voice communications are not sensitive to these shifts, nor are the typical tests run by telephone personnel. From the description of the DMS-100 problem, I would guess that the line interface circuit was missing or inserting extra samples. It is likely that synchronization problems will get worse rather than better, based on divestiture. Most of the following information is based on the publication Notes on the Network, 1980, AT&T, select code 500-029, Section 12. As digital trunks started to be deployed in the network, the need for a master clock to sychronize these systems was evident. AT&T built, and maintains a reference frequency which includes three cesium-beam frequency standards, accurate to within one part in 10^11. This clock is distributed in a strict hierarchy. This reference frequency is the strata 1 clock. It distributes this clock to strata 2 offices (for example, 4ESS (TM)). A 4ESS has a clock with a drift (in the absence of the strata 1 clock) of less than 1 part in 10^10 per day. During normal operations, the slip rate is zero (temperature fluctuations may cause occasional slips). During problems with connections to the strata 1 clock, the slip rate should be less than 1 in 10 hours. During a failure at the extreme design limit of sync problems, the number of slips should not exceed 255 per day, or about one "}" per 6 minutes. Note that this is considered a failure condition, not typical operating conditions. Now for the fun part. Where do telephone companies, such as MCI and Illinois Bell get their reference frequencies? Naturally, MCI does not wish to get their timing from AT&T. Illinois Bell typically would get its timing from its digital trunks to AT&T. The problem comes in when trying to enforce a timing hierarchy, even between local switches. It takes a considerable amount of engineering work to ensure this. The other problem is convincing someone that there is a problem in their network. One phrase that may help the telephone company isolating the problem is asking them to investigate the slip rate that the transmission equipment is reporting. It is unlikely that you can give this information to the person answering the repair phone, as in my experience, all they know about is that you will get charged if they find that the problem is in your wiring.
kaufman@SHASTA.STANFORD.EDU (Marc Kaufman) (10/25/87)
In article <KPETERSEN.12345088748.BABYL@SIMTEL20.ARPA> jhh@ihlpl.UUCP (Haller) writes: > ...give this information to the person answering the repair >phone, as in my experience, all they know about is that you will get >charged if they find that the problem is in your wiring. "It will be fixed tomorrow before 5 o'clock" My local office just cut over to ESS (from #5 crossbar). Evidently I can now buy a "data guaranteed" line for $$ extra per month. I have not been able to get a guarantee that if "I" have a data guaranteed line, and the other end (computer) has a data guaranteed line, ... that the central office trunks will preserve the data. I have just gone through a very long hassle with the local Telco over 48 InWATS lines that are used for data. The upshot is that "there is no data tariff for WATS", so they refuse to test for impulse noise and other forms of data degradation as long as the circuit can be used for voice. Thanks to deregulation, I discovered that if you get T1 service from the local toll office, you cannot get straight digital passthrough from the toll trunks. The lines are individually broken down to 2-wire analog circuits, sent through the nearest CO, and re-digitized for transmission to the customer! Fun, huh? Marc Kaufman (kaufman@Shasta.stanford.edu)