Exchange 2010 w/CAS Array (NLB) to Exchange 2013 migration

Hello,

This is an Exchange 2010 to Exchange 2013 migration question; how to introduce Exchange 2013 with existing Exchange 2010 CAS servers using NLB.

Currently my Exchange 2010 environment has 4 servers:
2 CAS/HUB (CAS Arrary using NLB)
2 MBX (DAG setup)

Goal is introduce 4 EX13 servers with High Availability (HA):
2 CAS (use NLB)
2 MBX  (use DAG)

The two EX10 CAS/HUB servers are part of a single CAS Array. High Availability is done using built-in Windows NLB, so both CAS/HUB servers are part of a single NLB cluster. The two MBX servers are part of single DAG.

I understand that EX13 no longer uses CAS arrays but rather OutlookAnywhere. My questions is how to introduce Exchange 2013 servers into the above environment and not break existing EX10. Installing the 2 EX13 MBX servers seems straight-forward, but cannot find guidance on how to introduce EX13 CAS servers when using NLB for HA. 

When configuring the two EX13 CAS servers with NLB installed, am I joining the new servers to the existing EX10 NLB cluster or am I creating a new NLB cluster (new IPs, etc) that will have only the two new EX13 CAS servers as members?

Thanks in advance.

June 8th, 2015 11:08am

The only thing you really have to do to avoid interference with newly-added servers is to enter the Set-ClientAccessServer command with the -AutodiscoverServiceInternalUri property to set the SCP to the address that you use for Autodiscover.  Otherwise clients will sometimes get the servername URL and throw a certificate warning.  Otherwise the Exchange 2013 server will sit there happily running and not bother anyone.

You do not join the Exchange 2013 servers to the Exchange 2010 NLB cluster.  Likewise you do not join them to the DAG.  Build the Exchange 2013 infrastructure, test that proxy and redirection works, then switch DNS to point the URLs to the new se

Free Windows Admin Tool Kit Click here and download it now
June 8th, 2015 7:59pm

Hi ,

Thank you for your question.

We suggest you use HLB in Exchange 2013. The following is disadvantage for DNS RR and WNLB:

DNS round robin does not provide persistent. It merely provides a means to alternate which servers are responding at a particular time. It does not provide any sort of persistence, or necessarily fault tolerance. So we dont recommend it.

Windows Network Load Balancing, and from a cross site architecture, while there are ways to deploy a single, unified, namespace across multiple datacenters, its extremely complex, it requires extremely expensive load balancing solutions you can do it, but it is an operational cost that you have to manage.

In addition Eds suggestion, we could refer to the following link to upgrade Exchange 2010 to Exchange 2013:

https://technet.microsoft.com/en-us/library/jj898583(v=exchg.150).aspx

In term of URL, we could refer to the following link:

http://blogs.technet.com/b/exchange/archive/2013/05/23/ambiguous-urls-and-their-effect-on-exchange-2010-to-exchange-2013-migrations.aspx

If there are any questions regarding this issue, please be free to let me know.

Best Regard,

Jim

June 9th, 2015 5:18am

Jim,

Thanks for the info and URLs; very useful. My org is not using ambiguous URLs; in that the CAS Arrary Object only points to an internal DNS object for MAPI/RPC and OutlookAnywhere points to an externally resolvable DNS object. Took a look at the various virtualdirectory settings and everything appears to be correctly separated (internal vs external). 

I have a small site where NLB with EX2010 works really well, so not worrying about a more complex cross-site config. Would like a L7 HLB but is cost prohibitive. 


Free Windows Admin Tool Kit Click here and download it now
June 10th, 2015 10:33am

Ed,

Thanks for the reply. Very useful to know that the EX13 setup should be in essence separate from existing EX10. Follow-up to that just to make sure I understand:

I plan on having 2 EX13 CAS servers using NLB. I should create a new NLB cluster with a unique IP and unique namespace? 
--Current EX10 CAS NLB Cluster name is set to the CAS Arrary Object value (resolvable via internal DNS only); lets say this value is:
A.internal.contoso.com

Should I set the EX13 CAS NLB name to something uqniue, i.e.:?
B.internal.contoso.com 

Also, thanks for the warning about AutodiscoverServiceInternalUrl property; will take a look to make sure its set correctly.  

June 10th, 2015 10:42am

You would need a new NLB cluster (HLB is better).  You will want to make the Exchange 2013 CAS servers your point of entry for services unless you want two separate URLs by version, which is inconvenient and confusing for users.

Free Windows Admin Tool Kit Click here and download it now
June 11th, 2015 1:18am

Jim,

Thanks for the info and URLs; very useful. My org is not using ambiguous URLs; in that the CAS Arrary Object only points to an internal DNS object for MAPI/RPC and OutlookAnywhere points to an externally resolvable DNS object. Took a look at the various virtualdirectory settings and everything appears to be correctly separated (internal vs external). 

I have a small site where NLB with EX2010 works really well, so not worrying about a more complex cross-site config. Would like a L7 HLB but is cost prohibitive. 


Hi Sheep,

If you think HLB is cost prohibitive, we could use NLB.

In addition Eds suggestion, we should make sure Exchange CAS server is separate from Exchange 2013 mailbox. Because WNLB is incompatible with Windows failover clustering. If you're using an Exchange 2013 DAG and you want to use WNLB, you need to have the Client Access server role and the Mailbox server role running on separate servers.

If there are any questions regarding this issue, please be free to let me know.

Best Regard,

Jim


June 11th, 2015 4:23am

Ed,

One additional question about your comment on setting the "AutodiscoverServiceInternalUri" property to the SCP of our autodiscover A record. 

I checked our existing EX10 CAS servers and they have the "AutodiscoverServiceInternalUri" set to their respective FQDNs:

https://CAS01FQDN.company.com/Autodiscover/Autodiscover.xml on one box
https://CASFQ02DN.company.com/Autodiscover/Autodiscover.xml on other

The EX10 unified certificate currently has the CAS01FQDN and CAS02FQND as SAN, so clients are not getting any cert warnings/errors. Cert SAN entries also have an internal only host name "autodiscover.a.b.c"

Initially will introduce two EX13 mailbox servers, then add two EX13 CAS servers. For your suggestion, do you mean to set the AutodiscoverServiceInternalUri property on EX13 CAS servers only to  
https://autodiscover.a.b.c/Autodiscover/Autodiscover.xml

Should this setting be changed on EX10 CAS servers?
Thanks in advance. This is basically a new question so let me know if I should start new thread.

Free Windows Admin Tool Kit Click here and download it now
June 17th, 2015 4:15pm

Set the AutodiscoverServiceInternalUri to the same value, the one that points to the Exchange 2013 server, on all Exchange CAS servers.
June 19th, 2015 2:20am

This topic is archived. No further replies will be accepted.

Other recent topics Other recent topics