[cisco-voip] building scalable dial plan

Wed May 9 10:12:21 EDT 2007

Hi Kevin,

I had a very informative discussion few months ago with Dan in this
group. I will just repost our conversation, about scalable dialplans.
Read from the bottom up!!

Ryan,

     -Yes each location has there own customized translation pattern to
expand a 4 digit call to a full 10 digit extension.

     - Funnny you should ask.  No I don't like to send the @ route to
the GateKeeper.  I like to be more specific.  I have been thinking of
adding that to the design but I would prefer to perform more testing.  I
would love to finally be able to bring up my missed call log, see a
missed call from an E164 number, hit the dial key and have it route the
call correctly.  The typical user doesn't like to see the prepended 91
in their call logs.  I agree. 

     -Wow you have put a lot of thought into your dialplan, good for
you.  I only configure zones for DID numbers.  The sites operate just as
you described.  I put non DID extensions into a seperate partition.
Then give them a fake E164 number beggining with a 1.  111.901.4567.
The remainder of the number can mirror the DID(s) for the site.  You
want to plan ahead to make sure you will never have overlap in Unity.
CallManager ois finished here.

     -Centralized Unity just works.  No translations, no masks.  The CSS
for Unity includes all SITE-DID partitons as well as SITE-NonDID
partitions.  There is no overlap so Unity is capable of sending MWI
On/Off as well as transferring to all extensions.  This also serves as a
back door transfer to non DID extensions. 

     -CAC is used here.  The LOCATIONS in CallManager are enforced and
you have additional bandwidth policing at the GateKeeper.

Everything just works!

On 3/12/07, Ryan O'Connell <Roconnell at unislumin.com> wrote: 

Dan,

I like it, very slick. Thank you for putting such detail in your
response as well. I hope you don't mind me asking a few more questions
I'm intrigued now. Besides I'm still early in the design phase and you
may sway me to alter my plans. 

-        In your design you have all your internal DN's set to be E.164
numbers. Are you using some sort of Translation patterns at each of the
sites to maintain 4 digit dial within a given site?

-        I don't have a lot of experience with Gatekeepers so let me ask
you this when using a Gatekeeper controlled trunk is it just one route
into CallManager like a default route? Using your example: User dials
91.6025028867, the RG-1 strips predot and sends it to Gatekeeper. I
assume there is a default route command that sends all 10 digit numbers
to CCM. Once it gets there since the Trunks Inbound CSS has all
"SITE_EXTENTIONS" partitions so therefore the call completes. 

-        What about non-DID's at remote sites do you have them in
another partition or something that the Gatekeeper can't see? Example at
your remote site you don't use DID's so do you still assign them a E.164
number? What happens if that E.164 number is an actual number for
MacDonalds at that remote location but it happens to be the same number
that you assigned to a non-DID at the remote location?

-        What about voicemail, I'm not sure if you're using centralized
voicemail with Unity but that is what I'm using. How are you handling
Trap connections and MWI for remote locations? Do those outbound Unity
ports have the same CSS that you applied to your Gatekeeper controlled
trunk? 

-        With your design are you still using CAC? 

Aren't Dialplans fun!!

Cheers Ryan

________________________________

From: Dan Cavanaugh [mailto:dcavanaugh at gmail.com]
Sent: Mon 3/12/2007 6:33 PM
To: Ryan O'Connell
Cc: Justin Steinberg; cisco-voip at puck.nether.net 

Subject: Re: [cisco-voip] MWI with overlapping numbers

The GateKeeper helps you perform a basic translation.  The way I have it
configured is each site on the network uses (2) route lists, SITE-RL and
SITE-GK-RL.  All local route patterns, local 7 digit, 911, etc. use the
SITE-RL which points to the local VG.  You are done there.  Any
91.[2-9]xx[2-9]xxxxxx dial pattern is routed to the SITE-GK-RL.  Keep in
mind that ALL your extensions at all sites are the full E164 number,
6025028867.  A user dials 91.6025028867, not knowing if this number is
on your network or a local McDonalds.  The SITE-GK-RL consists of (3)
route groups and does the following:

(RG-1)  Strip the preDot and send it to the GateKeeper.  The GateKeeper
will maintain ranges of extensions available on the CallManager cluster
or multiple clusters(60250288..).  The GateKeeper controlled trunks to
CallManager will have a CSS that has access to all SITE-EXTENSIONS
partitions.  If the 10 digit number matches a defined range on the
GateKeeper, the call is placed ON-Net.  If this request fails, --> RG-2.

(RG-2)  Send it to GateKeeper.  The GateKeeper will maintain a list of
91xxx....... routes.  The xxx will be the local area code for each
gateway registered to GateKeeper.  If the GateKeeper has a gateway
registered for 91602......., the call will be routed to a gateway on the
network and the call will be placed as a local call.  If there is not a
registered gateway with the 91602 zone defined, -->RG-3. 

(RG-3)  Finally send the call to the local VG as a long distance call.

You said you weren't doing any intercluster trunks.  The only trunk
needed would be to the GateKeeper.  It would take place of your CSS and
PT manipulations.  This is very modular and scalable in design.  It also
allows you to just plug in another site(or company) overnight without
further thought about overlapping dialplans.  Users don't need to be
notified about dialing rules for ON-Net vs. OFF-Net calls, they just
simply call each other the same way they always have.  What it does not
accomplish is the silly abbreviated dialing between sites.  Everyone
would love to dial across the nation using only 4 digits.  This is a
pipe dream and you will spend many nights undoing what you thought would
scale for your company.  

On 3/11/07, Ryan O'Connell <Roconnell at unislumin.com > wrote: 

Hi Justin

I would like to hear more about your comments you made about problems
with CTI and AC line status. How will this be a problem? 

Also Dan, why introduce a Gatekeeper how will this help on a centralized
design, I'm not doing any intercluster trunks? 

Ryan

________________________________

From: Dan Cavanaugh [mailto: dcavanaugh at gmail.com] 
Sent: Friday, March 09, 2007 10:26 AM 
To: Justin Steinberg
Cc: Ryan O'Connell; cisco-voip at puck.nether.net
Subject: Re: [cisco-voip] MWI with overlapping numbers

Total agreement with Justin.  What you have done at this point is
introduced layers of translation and complexity that don't need to be
there.  A nice design is exactly what Justin suggests below, with the
addition of a GateKeeper.  The Gatekeeper will add the filter you need
to determine if an extension is on your cluster. 

On 3/6/07, Justin Steinberg < jsteinberg at gmail.com
<mailto:jsteinberg at gmail.com> > wrote: 

You will notice other problems with this approach such as:

CDR issues since CDR is not partition aware 
CTI issues 
AC issues with line state updates

I think it would make life easier to use E164 numbers as the DN and use
translation patterns within each site to translate four digits to E164. 

Just my thoughts, and I do realize making this conversion would be a
real pain. :) 

Justin

On 3/6/07, Ryan O'Connell < Roconnell at unislumin.com
<mailto:Roconnell at unislumin.com> > wrote:

Fixed it

The fix was to change the MWI CSS to search the translation Table
instead of each sites local partition. And to change the MWI
notification in Unity back to "X" 

Ryan

________________________________

From: cisco-voip-bounces at puck.nether.net on behalf of Ryan O'Connell 
Sent: Tue 3/6/2007 3:07 PM
To: cisco-voip at puck.nether.net
Subject: [cisco-voip] MWI with overlapping numbers 

Hey all,

I'm stuck on this one. Here is the scenario.

IPT customer, with lots of phones. They are using a Centralized model
where CallManger and Unity are at the head office providing services for
head office and all branch locations. This is a big customer and DN
overlapp is a real issue so they decided they wanted to go 4 digit
dialing within a given site, and 10 digit dialing between sites. 

No problem

*	I setup each location with a different <sitename>Oncluster
Partition and added 4 digit DN's to these partitions. 
*	phones can dial each other using 4 digits within a site 
*	when they dial another site the must dial 11 digits. example to
dial Edmonton they dial 917808885180 
*	A translation pattern is used to strip the pattern and search
the correct sites partition then strip the 91780888 off the number
leaving 5180 
*	For centralized voicemail, I used different "VoiceMailProfiles"
to prefix the numbers going into Unity so when someone from Edmonton
dials into Unity 780888 is prefixed to the 5180 putting them into the
correct mailbox. 
*	To light MWI lamps, I had to turn Multitenant Mode on so that VM
Ports can see the translations created and all works fine 

HERE is the problem

*	When someone has the same last 4 digit dn's at different sites,
MWI only works for the site who's partition comes first in the MWI CSS  

Example Calgary phone with DN 4035555180 and an Edmonton Phone
7808885180

*	Unity mailbox are as follows Calgary user is 4035555180 and the
Edmonton User is 7808885180 
*	both message notification feilds under each subscriber have been
changed from the default X to 5180 because the actual extension is only
4 digits not 10 (I THINK HERE IS MY PROBLEM BUT DON'T KNOW HOW TO FIX
IT) 
*	I leave a message for the Edmonton user 
*	MWI outbound ports use the Translation pattern to translate
7808885180 to 5180 in the edmOnCluster partition. And tell the phone to
dial the lamp code 
*	The MWI-ON CSS has partitions in the order of cgyOncluster -
then edmOncluster. It doesn't know better it just matches 5180 to the
cgyOncluster and the lamp turns on. 

I know you may need a decoder ring to understand this email, but any
help is appreciated

Ryno

________________________________

________________________________

From: cisco-voip-bounces at puck.nether.net
[mailto:cisco-voip-bounces at puck.nether.net] On Behalf Of kevin k
Sent: Tuesday, May 08, 2007 10:09 PM
To: cisco-voip at puck.nether.net
Subject: [cisco-voip] building scalable dial plan

new to voip. reading and reading.

situation.

central ccm cluster.

i have over 200 sites.

5 sites - voip

rest centrex + key system

the debate is what to do with dial plan.

users at each site are used and fond of using 4 digit dn internally.

but looking ahead 4 digit dn will not suffice.

can you experts out there, suggest how to give 4 digit dn internally and
10 digit outside of the site, inter-site.????

or any thought?

i appreciate any comment.

Attention:

Privileged/Confidential Information may be contained in this message.  Disclosure to any person other than the named recipient is unauthorized.  If you are not the intended recipient, please delete all copies of this information and kindly notify the sender by reply email.  Opinions, conclusions and other information in this message that do not relate to the official business of UNIS LUMIN Inc. shall be understood as neither given nor endorsed by it.  UNIS LUMIN Inc. and any of its subsidiaries reserve the right to monitor all e-mail communications through its networks.  Thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://puck.nether.net/pipermail/cisco-voip/attachments/20070509/2b8c3908/attachment-0001.html 