Search This Blog

Friday, February 5, 2010

Unity - DR

This post would give a high end idea on how to configure Unity for DR, I would not go into a step-by-step installation and configuration,for these please refer the appropriate cisco documentation...

There are two scenarios to consider, first you have Unity in your environment and need to setup a secondary node configured for DR, second Unity is installed with failover and you need to shift one node to another datacenter to achieve site DR.



Scenario 1

Install Unity in DR datacenter,the IP address on this node would depend on whether you have stretch vlan's in your environement or not,Unity can be connected to a different subnet,Unlike windows clustering Unity does not have the requirement for both the nodes to be in the same subnet/vlan.



http://www.cisco.com/en/US/docs/voice_ip_comm/unity/42/upgrade/guide/ex/ru_090.html#wp1124930


Scenario 2

Failover Redundancy (Hot Standby) - If we have a LAN grade connection between the two sites then all we need to do is move the passive
node to the DR datacenter,

- Break the current failover config
- Move passive node to DR datacenter
- Power-up passive node and change the IP address (make sure DNS is
updated)
- Reconfigure failover
- Reconfigure the ports in Call Manager for node2 (Note: If Unity is
integrated with multiple CCM clusters,the changes need to be done on
all CCM clusters)


http://www.cisco.com/en/US/docs/voice_ip_comm/unity/42/upgrade/guide/ex/ru_090.html#wp1124930


If you have stretch vlan’s between the two Datacenters then the above is not required, just shutdown node 2 -> move to DR datacenter -> connect passive node to the same vlan and power up and that’s it, node2 would work as expected.


Standby Redundancy (Cold Standby) - Manual Intervention is required, the configuration is same, only automatic failover/failback is disabled,

http://www.cisco.com/en/US/docs/voice_ip_comm/unity/5x/requirements/50cusysreq.html#wp575709


The above listed scenarios assume that you have a tested Exchange DR in place. Now lets assume your primary datacenter goes down, Unity would immediately failover to the DR datacenter but manual reconfiguration might be required depending on how your Exchange DR is setup, If you are using Windows Clustering(Exch 2003) or CCR (Exch 2007) and Exchange services on the passive cluster on DR site start normally then ideally no change in Unity config w.r.t to exchange would be required, this is assuming Unity is pointing to the cluster hostname not the hostname of the exchange box.

The Call Manager failover to the passive Unity node should be automatic and no changes should be required on CCM, This would work in both scenarios, CCM hosted on the primary datacenters and atleast one CCM subscriber node on the DR datacenter or CCM hosted on some other datacenter, eg:- In a managed hosting environment CCM mightnot be colocated with Unity and might be homed at your clients datacenter.

Unity - Advanced Settings for 72+ Ports

By default Unity uses 1 MAPI connection to Exchange for a Unity system with 72 ports, If Unity has more than 72 ports and users complain of slowness in VM access (login to VM) or VM delivery to Exchange mailbox and the following event log is consistently observed in the App Event Log then you may look at changing the following settings:




On the Unity box goto to Tools Depot, expand the Administration Tools and double-click Advanced Settings Tool.In the left pane, click Messaging - 72 or More Voice Ports - Enable Low-Fragmentation Heap, In the New Value drop-down box, click 1 and click Set and when prompted click Ok,



Inthe Cisco Unity Advanced Settings window, in the left pane, click Messaging - 72 or More Voice Ports - Number of MAPI Sessions Per Exchange Server.

Number of VM Ports Setting
72 to 83 Click 2
84 to 95 Click 3
96 to 119 Click 4
120 to 143 Click 5
144 Click 6




When prompted that the value has been set, click Ok,Close Tools Depot and bounce Unity.

Note: If Exchange is experiencing high Disk I/O due to very high mailbox usage,under performing disk subsystem,migration/database replication applications reading writing to mailboxes/databases, then the above change will not make much difference as Exchange performance would be the root cause for slow VM access,The following Exchange perfmon counters will provide a good insight on Exchange performace,

Average Disk sec/Read - Average < 20ms, spikes < 50ms
Average Disk sec/Write - Average < 20ms, spikes < 50ms
RPC Averaged Latency - Always < 50ms

There are many counters to check Exchange Disk I/O but the above is suffice to judge Exchange performance.The below MS technet article gives a good insight on what causes Exchange Disk I/O:

http://technet.microsoft.com/en-us/library/aa996376(EXCHG.65).aspx

Viewmail for Outlook (VMO)

Viewmail for Outlook (VMO) Issue in Authentication-Resource Forest environment

VMO constantly prompts for authentication

The way VMO works is, it takes the credentials that were passed to it and then it needs to verify that if the user is a Unity subscriber so that it makes the correct integration to make the Outbound Call to CCM, If the user domain credentials are those of Authentication Forest (domain1) say domain1\user1 and Unity is installed in Resource Forest (domain2) then in the Unity database the user would be registered as domain2\user1 and the database lookup would fail.

The way to resolve this is to use the Unity tool grantunityaccess to make a database relationship between domain2\user1 and domain1\user1,

C:\commserver\grantunityaccess –u domain1\user1 –s user1

If the above fails try the following

C:\commserver\grantunityaccess –u user1@domain1.com –s user1

Also note that user1 in domain2 would be having a linked mailbox (in case of Exchange 2007), the exchange alias for user1 in domain2 should be used with the above command.

The above would work fine if you have a 2 way trust between the Authentication and Resource Forest.