cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Announcements

202
Views
5
Helpful
3
Replies
Beginner

Question about Deployment Model

We have an LA headquarters with 2 UCS servers. At the moment is it a centralized call processing. CUCM Pub/Sub and UnityConn Pub/Sub. We are bringing up a NYC branch and have a few options open to us. (currently we have 1 UCS server in NYC but its not being used for anything)

Given best practice deployment models, which is a better deployment ?

Option # 1 : Put a LA CUCM Subscriber in NYC which the NYC phones will use to register too but calls will come out of the NYC gateway via the PRI. This option will also include a UnityConn PUB in NYC and a UnityConn Subscriber in LA so that if the UCS server in NYC goes down, voicemail will be uneffected.

Option # 2 : Go full distributed call processing, LA is its own Site. NYC is its own site with its own CUCM Pub and Unity Pub. Gateway will have SRST for failover and there will be a NYC UnityConn Sub in LA for voicemail failover

what are the pro's and con's of these options ?

much appreciated , Thank you

1 ACCEPTED SOLUTION

Accepted Solutions
Hall of Fame Master

Question about Deployment Model

How big are the sites, i.e. how many users?

If you can limit number of cluster that is always preffered as you don't have to maintain 2 separate systems, worry about separate upgrade, feature parity between users, etc.

Splitting cluster geographically as listed in option 1 is always a good idea as long as you can meet the stringent requirements listed in SRND, i.e. latency below 80 ms, adequate bandwidth for DB replication and ICCS traffic (pretty significant), DB bandwidth for UCXN (7 mbps per 50 ports0. It also comes with considerations for your CUCM group building for redundancy, do you do local 1:1 model, 2:1, is the backup server at another location, etc, etc.

HTH,

Chris

3 REPLIES 3
Hall of Fame Master

Question about Deployment Model

How big are the sites, i.e. how many users?

If you can limit number of cluster that is always preffered as you don't have to maintain 2 separate systems, worry about separate upgrade, feature parity between users, etc.

Splitting cluster geographically as listed in option 1 is always a good idea as long as you can meet the stringent requirements listed in SRND, i.e. latency below 80 ms, adequate bandwidth for DB replication and ICCS traffic (pretty significant), DB bandwidth for UCXN (7 mbps per 50 ports0. It also comes with considerations for your CUCM group building for redundancy, do you do local 1:1 model, 2:1, is the backup server at another location, etc, etc.

HTH,

Chris

Highlighted
Beginner

Question about Deployment Model

The NYC site would be only about 10-15 users. You bring up a very good points on why to go with option 1 about splitting the cluster geographically . That makes more sense since it will only be a small site and we will have frame relay between the sites .

Thank you very much Chris,

Hall of Fame Master

Question about Deployment Model

Unless you want to build out redundant data center for DR having a dedicated sub for 15 users does not make sense, simply add SRST GW for failover and call it a day.

Chris

CreatePlease to create content
Content for Community-Ad
August's Community Spotlight Awards