|
 |
| Switch
Watchdog |
|
(Actual
screenshot of the Watchdog monitoring an Excel LNX gateway switch
to Southeast Asia) |
|
| Background: |
|
| Switch
Watchdog was developed by an international facilities based
carrier to monitor real time traffic statistics on their international
gateway switches. Prior to Watchdog, this carrier was
relying on trouble tickets and daily printed reports of the
previous day's traffic to discover problems on their inbound
and outbound routes. But this method was delivering the
news of problems many hours too late, and was causing their
customers to route away to alternate carriers. What the
carrier needed was a way to see problems before their customers
ever knew they existed... |
|
|
Overview: |
|
| Switch
Watchdog is a call statistics application that monitors switch
traffic in real time and sends out alarms when quality falls
below user-defined thresholds. The program receives CDR records
directly from the switch as each call is completed, and then
calculates statistics that affect route quality. When
any of these statistics drop below user-defined thresholds,
Switch Watchdog sends out alarms through email, the PC's sound
card, or to an optional PC relay card that ties into a switch
room's alarm board. In addition to its alarm functions,
Watchdog performs well as a switch activity monitor showing
up-to-the-minute traffic data in colorful charts and tables.
|
|
| Testimonials: |
|
| "Running
our switch without Watchdog is like driving a car without a
dashboard!" |
| --Ferdie
Yambot, CTO, Eagle Bell |
|
| "It's
how we stay on top of our outbound carriers and kick their butts
when their quality drops. They ask us how we knew, and we just
say Watchdog." |
| --Tarek
Jamal, CEO, NovaTel Global |
|
| "Watchdog
gives us immediate notification of network trouble so we no
longer need technicians on site 24/7." |
| --Ronald
Ramos, Technical Director, Innocom Telecom LLC |
|
| Features: |
|
Screen
Elements
Watchdog's screen has been carefully designed to give a "one
look" assessment of traffic volume and route performance on
a switch at any given time. Its green CDR area scrolls
with each new call recorded to the CDR file and shows call duration,
inbound and outbound trunk groups, and the dialed digits.
(This is a favorite feature of switch operators when test calls
are being placed during switch-to-switch testing, and for general
observation of traffic volume.) Just right of the CDR
display, the red Alarms area shows the latest warning, alarms,
and notifications.
Below the CDR and Alarms areas, the inbound and outbound Summary
Stats tables display trunkgroup-specific counters for six key
statistics:
- Attempted Calls (Att.)
- Completed Calls (Comp.)
- Call Completion Ratio (ASR)
- Total Minutes (Mins)
- Average Call Duration (Dur)
- Average Answer Delay (Delay)
The Summary Stats tables can be sorted in either ascending or
descending order on any column by clicking on the respective
column heading. Counters accumulate until reset by the
operator using the pull-down menus.
The bar charts show the ASR for each inbound and outbound
trunkgroup over the last 60 minutes and they are updated every
15 minutes. When the ASR for any trunk group falls below
a user-defined threshold (say, 35%) an alarm is raised, i.e.
ASR for
0005 was 26.9%. Chart
bars are displayed in red when the level is in alarm territory.
Alarms
Watchdog supports ten different alarm types, each of which
can be individually configured. The alarm types are:
- Global volume change%
- Global no calls minutes
- Outbound trunkgroup in use
- Trunkgroup-specific ASR
- Trunkgroup-specific volume change%
- Trunkgroup-specific no calls minutes
- Trunkgroup-specific answer delay
- Trunkgroup-specific consecutive incomplete calls
- Trunkgroup-specific low ALOC 
- Trunkgroup-specific high ALOC
A detailed description of each follows:
Global Volume Change Alarm
User sets a threshold in percent, say 80%. Watchdog tallies
all calls going through the switch in 15 minute intervals.
If the number of calls in the most recent interval is 80% higher
or lower than the number of calls in the previous 15 minute
interval, the alarm is raised. Example alarm: 12/1/99
4:08:02 PM: Global Volume Change: +268% in last 15 mins.
(47 calls vs 173 calls).
Global
No Calls Minutes Alarm
User sets a threshold, say 20 minutes. If Watchdog
detects a period of 20 or more minutes without a single call,
an alarm is raised. Example alarm: 12/1/99
4:08:02 PM:
GLOBAL No Calls in last 20 minutes! As soon as the next
call is detected, Watchdog follows up with another alarm:
12/1/99
4:38:02 PM: GLOBAL No Calls alarm cleared after 30 minutes.
Outbound
Trunkgroup in Use Alarm
User enters a trunk group ID, say "0000". This can be
an "overflow of last resort" route, or in the case of the Excel/ISOS
host (Exchange Plus), the indication of an unroutable call.
When Watchdog sees a call going to 0000, the alarm is raised:
12/3/99
10:23:26 PM: Outbound
Trunkgroup in use: 0000. To cut down on excessive emailed
alarms, Watchdog issues the initial alarm immediately, but then
limits subsequent alarms to one every five minutes. These
"delayed" alarms also include the number of calls to 0000 that
occurred in that five minute interval. Example alarm:
12/3/99
10:43:26 PM: 17 route attempt(s) to Outbound Trunkgroup 0000.
Trunkgroup-specific Alarms
Trunkgroup-specific alarms can be set on any number of inbound
or outbound trunkgroups. They include alarms for low ASR,
volume changes, no traffic, and excessive answer delay.
Trunkgroup-specific ASR Alarm
A threshold is set, say 40%, and the call completion ratio over
the past 60 minutes is calculated each 15 minutes. If
it falls below 40%, the alarm is raised:
12/3/99 11:04:57 PM: ASR for O0010 was 21.5%.
Trunkgroup-specific
Volume Change Alarm
Similar to the Global Volume change described above, the trunkgroup-specific
alarm tallies all calls going through a trunkgroup in one direction
in 15 minute intervals. If the number of calls in the
most recent interval is the specified percentage higher or lower
than the number of calls in the previous 15 minute interval,
the alarm is raised. Example alarm: 12/3/99
7:15:06 PM: Volume Change on O0010: +248% in last 15 mins.
(50 calls vs 174 calls).
Trunkgroup-specific
No Calls Alarm
Similar to the Global No Calls alarm described above, the trunkgroup-specific
alarm is raised if no calls pass on the particular trunkgroup
for the specified number of minutes (i.e. 20 minutes).
Example alarm: 12/4/99
10:12:49 PM: No Calls in last 20 minutes for TG O0010!
Trunkgroup-specific
Answer Delay Alarm
Related to post-dial delay (PDD), answer delay measures the
interval between when the switch began handling the call, and
when the far-end party answers the call. Averaged over
many calls, the answer delay gives a very useful index into
PDD on the circuit, i.e. if the answer delay is normally 18-20
seconds, but suddenly is in the high-20s, something has likely
happened on the circuit that is extending PDD. The answer
delay alarm is raised when the observed answer delay exceeds
the threshold set by the user, say 22 seconds. Example alarm:
12/3/99
11:20:26 PM: Answer Delay for O0010 was 23 seconds.
Trunkgroup-specific Consecutive
Incomplete Calls Alarm
Tallies consecutive incomplete calls on a trunk group and raises
an alarm when the count rises above the threshold (i.e. 25 calls).
If the condition persists, alarms are repeated at same
interval, i.e. 50 calls, 75 calls, etc. Example alarm:
2/5/00 12:13:57
PM: 30
consecutive incomplete calls on I8750. As soon as
a completed call is detected on that trunk group, Watchdog follows
up with another alarm: 2/5/00
12:25:06 PM: Consecutive incomplete calls alarm for I8750 cleared
after 37 calls.
Trunkgroup-specific Low
ALOC Alarm
Measures average length of call (ALOC) and alarms when the ALOC
drops below the threshold. Example alarm: 2/5/00
12:25:06 PM: Low ALOC: 27 seconds on TG I0230 over past 60 mins!
(385 calls)
Trunkgroup-specific High
ALOC Alarm
Measures average length of call (ALOC) and alarms when the ALOC
exceeds the threshold. Example alarm: 2/5/00
12:25:06 PM: High ALOC: 1204 seconds on TG I3415 over past 60
mins! (145 calls)
Alarm
Outputs
Each of Watchdog's alarms can be individually set to go to any
of three outputs:
- E-mail
- Sound card
- Relay card
E-mail is the most common method of dispatching alarms from
Watchdog, and it's particularly useful for "tech-on-call" personnel
and during hours when your site is unmanned. It requires
a full-time Internet connection so Watchdog's built-in SMTP
client can send emails to your mail server. Watchdog's
alarm messages are kept brief to allow the use of email-to-pager
gateway services that let you receive e-mail on an alphanumeric
pager.
On-site personnel can be informed of alarms by either the sound
card or relay card alarm outputs. Using a standard "wav"
audio file of the user's choosing, any of Watchdog's alarms
can be configured to play the audio file through the PC's speakers.
The wav file is played continuously until the alarm is cleared
with the F12 key.
For switch rooms equipped with alarm panels, Watchdog's optional
ISA or PCI relay card supports normally-open and normally-closed
alarm interfaces. Up to eight separate Watchdog alarms
can be piped into your existing alarm board. ( Relay
card support is under development... call for scheduled release
date.) |
|
|
Architecture: |
|
Switch
Watchdog is a client/server application where the Watchdog client
runs on Windows 95/98/NT/2000 and receives CDR data on a TCP/IP
socket from the switch host server. This IP connection
can link Watchdog to a switch in the same room using a LAN,
or Watchdog can monitor a switch in another country using the
Internet.
Some switches have the built-in ability to broadcast CDR information
on a socket. Others require an additional program that
Switch Management developed called cdrserver.
Cdrserver is a Unix/C program that spies on the CDR data file
as the switch is adding to it on disk. As new rows are
added to the file, cdrserver broadcasts them out its TCP/IP
socket. Firewall safeguards are in place to limit connections
to clients with known IP addresses. |
|
| Supported
Telecom Switches: |
|
Watchdog
currently supports the Excel, Clarent, Summa Four, NACT, and
Aculab switches and these host control programs:
- Clarent Command Center (SQL7 or Oracle)
- Excel/ISOS
- Excel/Exchange PLUS
- Excel/SoftSwitch
- Excel/XNT ADS
- Excel/Proactive GlobalCall
- Excel/Phoenix (Telemedia) CMS/2000
- Excel/iSoftel
- Siemens-Stromberg DCO
- NACT STX
- Summa Four/Compro Technologies
- Aculab Groomer
Don't see your switch in this list? Switch Management
is interested in talking to you about porting the Watchdog to
your switch. Contact our R&D department for more details. |
|
| Hardware
Requirements: |
|
Switch
Watchdog runs best on Windows NT Workstation 4.0, but Windows
95, 98, or 2000 will also work. Hardware requirements
are typical of current Windows applications:
- Pentium II or better
- 128MB Ram
- 4GB Hard disk
- Network card
Pricing:
Contact
the sales department for a quote at +1 925-283-9600 x2, or
send us your contact information on-line.
|
|
|
|
|
|
