ITISEMEA
asked on
Exchange 2010 CAS Array two active sites
Setup: Exchange 2010 SP3, windows 2008 R2 SP1
Data Centre
mbx01 - mailbox server
mbx02 - mailbox server
CAS01 - CAS/HT server
CAS02 - CAS/HT server
Head office
mbx03 - mailbox server
CAS03 - CAS/HT server
All active databases will be located at the Data Centre and a passive copy at the head office apart from head office users who's active database will be local and passive at the Data Centre
An F5 hardware loadbalancer is located at the Data Center and configured with CAS01 and CAS02 as the primary connections and if they are not available CAS03 as the tertiary.
Is this an acceptable configuration or shall I have all databases located at the Data Centre and head office as a DR site only with no active databases. Would make it easier in a DR situation.
How do I ensure that head office users connect to the local CAS Array instead of going across the WAN or does Outlook always look at the local site CAS first?
In the event we lose one site how will the failover occur?
Data Centre
mbx01 - mailbox server
mbx02 - mailbox server
CAS01 - CAS/HT server
CAS02 - CAS/HT server
Head office
mbx03 - mailbox server
CAS03 - CAS/HT server
All active databases will be located at the Data Centre and a passive copy at the head office apart from head office users who's active database will be local and passive at the Data Centre
An F5 hardware loadbalancer is located at the Data Center and configured with CAS01 and CAS02 as the primary connections and if they are not available CAS03 as the tertiary.
Is this an acceptable configuration or shall I have all databases located at the Data Centre and head office as a DR site only with no active databases. Would make it easier in a DR situation.
How do I ensure that head office users connect to the local CAS Array instead of going across the WAN or does Outlook always look at the local site CAS first?
In the event we lose one site how will the failover occur?
ASKER
Hi Sembee2
thanks for the quick reply. we have 1700 mailboxes.
makes sense regarding the site fail over and quorum.
90% of users will be accessing the mailboxes via the Data Centre therefore the head office will be the DR site.
considering i have an F5 in the scenario how would you suggest i configure the CAS array.
Data Centre - casrarray1.domain.local
head office casarray2.domain.local
DNS records for each name in AD
F5 pointing primarily to casarray1.domain.local and if they are not available point to casarray2.domain.local. in a DR event i would then follow the normal Exchange failover procedures?
thanks for the quick reply. we have 1700 mailboxes.
makes sense regarding the site fail over and quorum.
90% of users will be accessing the mailboxes via the Data Centre therefore the head office will be the DR site.
considering i have an F5 in the scenario how would you suggest i configure the CAS array.
Data Centre - casrarray1.domain.local
head office casarray2.domain.local
DNS records for each name in AD
F5 pointing primarily to casarray1.domain.local and if they are not available point to casarray2.domain.local. in a DR event i would then follow the normal Exchange failover procedures?
The RPC CAS Array is AD site specific.
Therefore as you have two sites you will need two CAS Arrays. In the event of a failover you would just move the DNS entry across to the other location.
Therefore the load balancer would just point everything to the server in the second location. The address the users are using to access Exchange with Outlook would not change.
Even with 1700 users I would still do the design I have outlined. I have just implemented (went live yesterday) 2500 users on the same design and it has worked exactly as planned.
You can even use the passive server in each location as the FSW for the active DAG, therefore keeping everything within Exchange.
Simon.
Therefore as you have two sites you will need two CAS Arrays. In the event of a failover you would just move the DNS entry across to the other location.
Therefore the load balancer would just point everything to the server in the second location. The address the users are using to access Exchange with Outlook would not change.
Even with 1700 users I would still do the design I have outlined. I have just implemented (went live yesterday) 2500 users on the same design and it has worked exactly as planned.
You can even use the passive server in each location as the FSW for the active DAG, therefore keeping everything within Exchange.
Simon.
ASKER
im still not certain about this F5 HLB configuration
i have a virtual IP ( 10.10.10.10) from the F5 which points to CAS01 and CAS01 primarily and CAS03 if 01 and 02 are down.
A DNS record has been created called CASArray1.domain.local which point to 10.10.10.10.
therefore if i also name the head office CASArray1.domain.local will the F5 not load balance this? Can you actually have the same name for a CASArray in different sites, will Exchange allow you do this.
i have a virtual IP ( 10.10.10.10) from the F5 which points to CAS01 and CAS01 primarily and CAS03 if 01 and 02 are down.
A DNS record has been created called CASArray1.domain.local which point to 10.10.10.10.
therefore if i also name the head office CASArray1.domain.local will the F5 not load balance this? Can you actually have the same name for a CASArray in different sites, will Exchange allow you do this.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Hi Simon
I just wanted to give you an update. i tried creating a new CASArray in the head office using the same name as the Data Centre and received the following error
Active Directory operation failed on GBLON01ADC03.domain.local.
ays,CN=Exchange Administrative Group (FYDIBOHF23SPDLT),CN=Admin
ervices,CN=Configuration,D
+ CategoryInfo : NotSpecified: (0:Int32) [New-ClientAccessArray], ADObjectAlreadyExistsExcep
+ FullyQualifiedErrorId : D947E4B1,Microsoft.Exchang
I just wanted to give you an update. i tried creating a new CASArray in the head office using the same name as the Data Centre and received the following error
Active Directory operation failed on GBLON01ADC03.domain.local.
ays,CN=Exchange Administrative Group (FYDIBOHF23SPDLT),CN=Admin
ervices,CN=Configuration,D
+ CategoryInfo : NotSpecified: (0:Int32) [New-ClientAccessArray], ADObjectAlreadyExistsExcep
+ FullyQualifiedErrorId : D947E4B1,Microsoft.Exchang
That is what I would expect. You cannot have two arrays with the same name. They must be unique. However you can, in the event of a failover, point the two arrays at the same CAS role holder.
Simon.
Simon.
ASKER
sorry i must have miss understood your previous post. Does the name need to be unique throughout the exchange environment, forest or just domain. We have several exchange servers in the enviroment which are split accross subdomains
The name and host name of the CAS array needs to be unique across the entire forest.
Where the Exchange servers physically are doesn't matter, as long as you have your AD Sites and Services setup correctly. If you have multiple subnets then you need to ensure that all of those are listed in AD sites and services and assigned to the correct site.
When I do multiple site implementations, we usually come up with a naming convention for the CAS array DNS entry.
outlook.eu.domain.local
outlook.na.domain.local
outlook.as.domain.local
etc
It makes it very easy to manage and know what host is for which location.
Simon.
Where the Exchange servers physically are doesn't matter, as long as you have your AD Sites and Services setup correctly. If you have multiple subnets then you need to ensure that all of those are listed in AD sites and services and assigned to the correct site.
When I do multiple site implementations, we usually come up with a naming convention for the CAS array DNS entry.
outlook.eu.domain.local
outlook.na.domain.local
outlook.as.domain.local
etc
It makes it very easy to manage and know what host is for which location.
Simon.
Personally I would redesign the platform to have all three roles on all the servers.
Have two DAGs, using three servers for each DAG. Two active in each site, with a single passive in the other. Each location would have its own CAS Array.
Remember CAS/Hub roles do not belong to any mailbox server, so the server can be used by users of DAG 1 while holding a database from DAG 2.
You haven't said how many users are involved, but I haven't done a design for less than 10,000 users in the last 18 months with seperate Mailbox and CAS/Hub.
Simon.