nanog mailing list archives
Re: DS-3 Error Stats
From: Sean Donelan <SEAN () SDG DRA COM>
Date: Mon, 14 Jun 1999 17:37:33 -0500
jhawk () bbnplanet COM (John Hawkinson) writes:
Most carries have "performance objectives" which they adhere to, and they specify some number of errored seconds per day per circuit, and those numbers may vary based on route-milage. Unfortunately I think most of those numbers are covered under NDAs, but I can safely say that 1 errored second/day would be well under the criteria and 1,000 would be well over, and anything in between depends on your carrier and route miles, etc., etc.
In most cases, DS3's are sold by carriers as a tariffed service. The carrier may not like disclosing the numbers, but buried in the public tariff's "incorporated by reference" list of documents are the performance objective for that carrier's DS3 service. Most carriers' tariffs refer back to either Bellcore(telcordia), ANSI/T1 or ITU specifications (generally in that order). However, I'm not sure knowing them will do you much good. I tracked down the numbers for DS1 once upon a time. Like any good-one-sided telco tariff, the circuit could be very bad and still meet the "acceptable" performance objectives for the service. If I recall correctly, a DS1 could have several thousand error seconds a day, and be considered "acceptable" according to the tariff. I've never looked up the DS3 numbers, so I don't know the official tariff numbers. As always, the tariff overrides anything the sales person told you, or anything written in your contract. However after saying that, digital lines usually only have 'noise' for a reason. The question is how bad does it have to be before you can get the telco to track it down. Here are my rules of thumb for most digital circuits: Error seconds/15 minutes Classification 0 Perfect 1-5 Look at the circuit during the next maintenance window if it doesn't clear up 6-29 Open a ticket 30+ Immediate action required (its likely the circuit is bouncing or 'down' at this point) Other people prefer to use a long-term bit error-rate, or packet/cell errors. But then you have to explain to the telco what a packet or bit is. For those ISPs which own their facility provider, they may be able to get the facility folks to monitor the circuits. But for the rest of us, you have to monitor your own circuits. The facility provider is never going to tell you about a potential problem. -- Sean Donelan, Data Research Associates, Inc, St. Louis, MO Affiliation given for identification not representation
Current thread:
- DS-3 Error Stats scott w (Jun 13)
- <Possible follow-ups>
- Re: DS-3 Error Stats scott w (Jun 13)
- Re: DS-3 Error Stats Alex Bligh (Jun 14)
- Re: DS-3 Error Stats John Hawkinson (Jun 14)
- Re: DS-3 Error Stats Alex Bligh (Jun 14)
- Re: DS-3 Error Stats Sean Donelan (Jun 14)
- Re: DS-3 Error Stats Mark A. Cnota (Jun 14)
- Re: DS-3 Error Stats Sean Donelan (Jun 14)
- Re: DS-3 Error Stats Mike Trest (Jun 15)