Re: (ASCEND) Ascend Disconnect Performance

[Mitchell Arnone <mitchell.arnone@occ.treas.gov>]

Thanks Will

98% of my users are Windows 95 but we do have some NT users and they 
consistently disconnect with a disconnect of 11 and progress of 60.

We do have Navis but it is only capturing disconnect information from the 
Ascend equipment.  We have some other non-ascend equipment as well so what 
I do is map all our NAS devices to the same RADIUS accounting server.  I 
wrote a program that decyphers the RADIUS logs then pumps the data to a SQL 
server.  From there I have writted stored procedures on SQL and developed a 
Web site with Active Server pages which run the stored procedures and 
presents the resulting information back to the Web interface.   The types 
of information available via the Web are ALL DISCONNECTS for a given period,
 ANALOG DISCONNECTS for a given period, USER STATISTICS searchable by user 
name, ALL USER ACTIVITY for a given period, USERS DENIED ACCESS for a given 
period showing the login name used and reason for the failed login, and 
canned reports like DATA RATE averages by day, TOP 20 LOGINS PERMITTED AND 
DENIED by day and MONTHLY DISCONNECT AVERAGES by month and day.

All this was custom written and we compare the results with reports from 
Navis Access.

Mitch
-------------
Original Text
From: "willp2" <willp2@dreamscape.com>, on 3/18/99 8:08 PM:
To: SMTP@DC2@OCC["Ascend Mailing List" <ascend-users@bungi.com>]

Running 40xx's here.  We've found that some users who run Windows NT ALWAYS
disconnect with a cause code of 11 (Lost Carrier) and a progress code of 60
(lan session up).  Some NT users always disconnect (every single time) with
a cause code 185 (Hung up) and progress code of 60.

It seems that the most consistent platform for generating a cause code of 
45
(User requested disconnect) is linux's pppd and win95/98.

If your customer base is primarily Windows NT, this will make a dramatic
difference in your statistics on disconnect reasons.  The real interesting
disconnect code that has helped our tech support department is the cause
code 47 (No NCP's opened).  This happens with corrupted windows registries
and can usually be fixed by uninstalling DUN and reinstalling it and
creating a new DUN connection (worst case scenario).

One thing we do to keep tabs on disconnect reasons is to keep a log of the
relative percentages of each disconnect code for each POP over the last 24
hours.  We've been able to find major facilities problems this way.  At
various times, I'll also run a report on disconnect cause codes for each
T1-span we have and scan them visually to see if any T-spans show an 
unusual
deviation.  I'm still trying to figure out how best to represent this data
so I can graph it with something like MRTG.  If anyone has any ideas on how
to visualize this data, speak up...  The best thing in the world was to be
able to find that a cross-connect cable had gone bad over a weekend by
seeing a 10% increase in cause code 11's (lost carrier) on a T-span.

-Will Pierce


-----Original Message-----
From: Cyril Jaouich <twiggy@twiggy.spider.org>
To: Mitchell Arnone <mitchell.arnone@occ.treas.gov>
Cc: ascend-users@bungi.com <ascend-users@bungi.com>
Date: Thursday, March 18, 1999 4:13 PM
Subject: Re: (ASCEND) Ascend Disconnect Performance


> Well, we don't use TNTs but with about 150 Max4048, the
>combination c=185 & p=31 plus c=10 & p=31 account for 3.5% of all
>disconnected calls.
>
>Cyril Jaouich [CJ837]


++ Ascend Users Mailing List ++
To unsubscribe:	send unsubscribe to ascend-users-request@bungi.com
To get FAQ'd:	<http://www.nealis.net/ascend/faq>

++ Ascend Users Mailing List ++
To unsubscribe:	send unsubscribe to ascend-users-request@bungi.com
To get FAQ'd:	<http://www.nealis.net/ascend/faq>