Hi

Thanks for your valuable input, I would like something

SITE1

UNITY1A (ACTIVE SERVER)

EXCHANGE1A

EXCHANGE2B

UNITY2A (FAILOVER SERVER)

AD,DC, GKS

SITE2

UNITY1B (FAILOVER SERVER)

EXCHANGE1B

EXCHANGE2A

UNITY2B(ACTIVE SERVER)

AD,DC,UNC

The idea is if one site loose connectivity, the user could go to disaster recovery site and continue working. How would you recommand to split user across two unity servers and how unity would know which user exists on which server.

thanks.

You can split the Primary and Failover server IF you following this:

Requirements for Spanning a Cisco Unity Failover Pair Across Redundant Data Centers

Revised May 6, 2008

Some customers maintain redundant data centers that are so well connected that they essentially form a single LAN. For these customers, failover may be a more suitable configuration than standby redundancy. For detailed requirements for this configuration, see the section "Failover Requirements for Separating Cisco Unity Servers by a WAN" in the System Requirements for Cisco Unity at http://www.cisco.com/en/US/products/sw/voicesw/ps2237/prod_installation_guides_list.html.

Standby Redundancy

Note the following about a Cisco Unity system configured for standby redundancy:

•The standby redundancy configuration requires a minimum available bandwidth of 100 Mb/second between the data centers in which the Cisco Unity primary and secondary servers are installed. This is because of two bandwidth spikes, one caused by SQL Server, the other by Cisco Unity:

-Once daily (by default at 2:00 a.m.), SQL Server replicates data from the active server to the inactive server.

-Cisco Unity periodically replicates WAV files of voice names, greetings, call handlers, and interview handlers from the active server to the inactive server.

•If something happens to the data center in which the Cisco Unity server is installed, the customer must have:

-A way to reroute call traffic to the data center in which the secondary server is installed.

-Sufficient bandwidth for call traffic to the data center in which the secondary server is installed.

•Automatic failover must be disabled when Cisco Unity standby redundancy is configured. Therefore, if the primary server becomes unavailable, the customer must manually make the secondary server active before it begins taking calls.

For a list of the requirements for standby redundancy, see the "Requirements for Standby Redundancy" section in the System Requirements for Cisco Unity Release 5.0 at http://www.cisco.com/en/US/products/sw/voicesw/ps2237/prod_installation_guides_list.html.

How do you know which subscriber is on which server, that is up to you. Like I said before you have to come up with a scheme for administration of 7000 subscribers. Somehow divide the subscribers between the clusters so you are balanced between the unity clusters. 3500 subscribers on one cluster, and 3500 on another cluster. There is no way to have 3500 subscribers from Cluster A to failover to cluster B. Unity does not work this way. It will only failover to the partner server. If no partner server is NOT available, you have no voicemail system for those 3500 subscribers.

Unity 7 is introducing some new concepts which I have not test yet. But may be worth the read i the SRND guides.

So, I will have one pilot number for voice mail for both clusters or two? In such case, I would need translation pattern to match to main pilot number. For redundancy, failover license works or there is special license for it.

thanks

Nope, you will have voicemail pilot for EACH Unity Primary and Failover pair.

Example:

Cluster "east coast"

Unity Server A1

Unity Failover B1

Cluster "West coast"

Unity Sever A2

Unity Failover B2

In CUCM, you will configure a pilot number "East Coast" and for "west coast" Each pilot number is associated to the hunt group which contains the voicemail ports for each of the cluster.

East Coast may have 144 ports with one Voicemail pilot number x5000. In the hunt group, 144 ports associated to East Coast cluster. Voicemail Profile Name "East Coast"

West Coast may have 144 ports with one voicemail pilot number x6000. In the hunt group, 144 ports associated to the West Coast cluster. Voicemail profile name "West Coast"

Make sense?

yup, but would like to have one main number for voice mail as user moves from one place to another and would like to have the feel of one voice mail system.

That would be a little tricky. You can network the Unity serves together and that will give them the "one number" feel. They can cross login into either cluster and retrieve their voicemail messages. The access numbers may be different because when the press the "messages" button, it will dial the voicemail pilot number for that cluster.

I was thinking along the cross server login. Same voice mail number for both clusters and user call handler and call routing on unity for second cluster.

Do you any high level design showing different elements such as AD,DC,GC,EX2K3 etc as to visualize the whole thing if possible.

I will be building vmachines with two unity clusters and testing it.

thanks

You still need to have (2) Pilot number. There is now way to have "1" number for them to dial. Either number will get them to the Unity server. Externally, you can have 800-xxx-xxxx to dial into Unity if you like.

But for the "messages" button on the phones, you have to setup separate pilot numbers. Otherwise you will basically have 7000 subscribers on one server.