Hi there,

I've just deployed a fresh CAS and setup a CAS Array, a second CAS will be put in with the next few days once I've sorted the main config out.

I've been going through sorting out the URLs and asking myself a few questions. Can or Should I be using the CASArray name in some of the URLs instead of the CAS host name. For example :

On the "Set-WebServicesVirtualDirectory -InternalUrl" im using https://CasArray.domain.com/ews....."

however on the "Set-ClientAccesServer - AutoDiscoverServiceInternalUri" im using https://CasServer1.domain.com/autodiscover...

Same for "Set-OabVirtualDirecotry -InternalUrl" = https://CasServer1.domain.com/OAB"

Also "Set-AutodiscoverVirtualDirectory -InternalUrl" = https://CasServer1.domain.com/autodiscover...

So my question is should I be using the Array name for any of these URLs or is that only used for the EWS InternalUrl

Any help here would be great.

Regards

(there wasn't a 2010 option for thread)

Hi

Actually I wouldn't use the CAS array name for any of the HTTPS workloads, while this is supported it is not recommend.  Rather create a DNS name like mail.yourdomain.com and give it the IP of your existing CAS server, use that name in all your HTTPS URLs.  When your second CAS is installed create another DNS record with the same name but give it the second CAS's IP address.

What you have now is a DNS round-robin for your CAS servers which is the simplest form of load balancing.  Should one of your CASs fail then you just delete the mail.yourdomain.com record for its IP address and the traffic will go to the working server.  Obviously this is pretty manual and you should consider a load balancing appliance of some sort but it will work.

You CAS array object and its name are only to be used for internal Outlook connections and nothing else.

Great blog series on this topic here: http://blogs.technet.com/b/exchange/archive/2012/03/23/demystifying-the-cas-array-object-part-1.aspx

More here (although this is more migration focused it does have some good background info): http://blogs.technet.com/b/exchange/archive/2013/05/23/ambiguous-urls-and-their-effect-on-exchange-2010-to-exchange-2013-migrations.aspx

Steve

Hi

Actually I wouldn't use the CAS array name for any of the HTTPS workloads, while this is supported it is not recommend.  Rather create a DNS name like mail.yourdomain.com and give it the IP of your existing CAS server, use that name in all your HTTPS URLs.  When your second CAS is installed create another DNS record with the same name but give it the second CAS's IP address.

What you have now is a DNS round-robin for your CAS servers which is the simplest form of load balancing.  Should one of your CASs fail then you just delete the mail.yourdomain.com record for its IP address and the traffic will go to the working server.  Obviously this is pretty manual and you should consider a load balancing appliance of some sort but it will work.

You CAS array object and its name are only to be used for internal Outlook connections and nothing else.

Great blog series on this topic here: http://blogs.technet.com/b/exchange/archive/2012/03/23/demystifying-the-cas-array-object-part-1.aspx

More here (although this is more migration focused it does have some good background info): http://blogs.technet.com/b/exchange/archive/2013/05/23/ambiguous-urls-and-their-effect-on-exchange-2010-to-exchange-2013-migrations.aspx

Steve

• Proposed as answer by Angela ShiMicrosoft contingent staff 1 hour 1 minutes ago

Hi

Actually I wouldn't use the CAS array name for any of the HTTPS workloads, while this is supported it is not recommend.  Rather create a DNS name like mail.yourdomain.com and give it the IP of your existing CAS server, use that name in all your HTTPS URLs.  When your second CAS is installed create another DNS record with the same name but give it the second CAS's IP address.

What you have now is a DNS round-robin for your CAS servers which is the simplest form of load balancing.  Should one of your CASs fail then you just delete the mail.yourdomain.com record for its IP address and the traffic will go to the working server.  Obviously this is pretty manual and you should consider a load balancing appliance of some sort but it will work.

You CAS array object and its name are only to be used for internal Outlook connections and nothing else.

Great blog series on this topic here: http://blogs.technet.com/b/exchange/archive/2012/03/23/demystifying-the-cas-array-object-part-1.aspx

More here (although this is more migration focused it does have some good background info): http://blogs.technet.com/b/exchange/archive/2013/05/23/ambiguous-urls-and-their-effect-on-exchange-2010-to-exchange-2013-migrations.aspx

Steve

• Proposed as answer by Angela ShiMicrosoft contingent staff, Moderator Tuesday, June 24, 2014 9:53 AM

Hi Steve, thanks for the reply.

We already have another CASArray with 2 CAS behind a WNLB, we were going to replicate this setup as its works very well for us (no cash for a hardware load balancer!).

If we do go with the 2xCAS\Array + WNLB what are the drawbacks of using the Array name in the https urls? Why is it not recommended.

Should I change my "WebServicesVirtualDirectory" back to CAS server? If I don't what bad things could happen..

Thanks again for your help.

Hi,

Firstly, I'd like to explain, it won't bring issue if you set the URLs with CAS array name.

However, here are the reasons which we don't use the CAS array name in the web services URLs:

1. CAS array is just used for MAPI connection and it has no effect on HTTPS connection.

2. Generally, we don't publish the CAS array name to the internet and add the name in the certificate.

For more information, you can refer to the articles that Steve provides above.

Thanks,

Ok - thanks to both of you for information on this.