Exchange 2010 DAG

You can give us more information about your plan in order to give you more detailled ideas about some recommended configurations

We know that you are considering to make a Geographic DAG including two sites and at least to Exchange 2010 servers. Give us these additional information: Complete list of the Exchange 2010 servers in each site and the roles that you plan to install in each one (CAS, HUB, Mailbox, etc)

Another extra recommendation if possible. Is to use a different connection for the repplication logs. In order to have more availability and less impact on your public network

Here's my config:
Two site with the same config: Three server each site
Server 1 with:CAS/HT,  Server  2 with CAS, Server 3 with Mailbox
Configure CasArray with Server 1 & 2
Configure DAG with Server 3 for both site

All Servers with two nic card installed.

Deal...

Like discussed before. Let's focus on one site, let's say that it is site A. If you make a CAS array with servers 1 and 2 it means that your MAPI clients will point to the array, in previous versions of Exchange the mapi clients connected to the mailbox server or back end where are the mailbox databases, but actually they connect to the CAS servers role, in this case your CAS array. With this you will be able to keep your users connected to the CAS array in case of a failover and you can have a delay of less than one minute or very few minutes in order to have your pointers updated after a failover or a movement of the resources

Some considerations:

- You actually have two nics on your mailbox servers. Use a dedicated network for the logs replication, in your case it can be the ones without gateway

- Configure your Cluster Failover / DAG to use a Fire Share Witness as a Quorum. It can be any server, but usually it is used an Exchange server because it usually means that are under the control of the same personal and the dependencies or problems are totally administered by this administrative / human group. Bas on this condition you can use the fire share witness on your server 1 or your server 2. Be aware that in order to bring service you need two of these three components: Fire Share Witness, Mailbox DAG member 1, Mailbox DAG member 2 (in the remote site). Because you are configuring a geographic cluster it is not a very design plan to accept that after losing the site 1 you lost your Fire Share Witness and one of your nodes and you can not start up your service. So... consider using also alternative Fire Share Witness server, in this case the server 1 or server 2 in the Site B

- Remember that your databases will be mounted and published for the CAS servers with one of these IP address according to the public IP address of your servers: 192.168.1.10 or 192.168.2.10. Because they are different IPs your CAS servers will have in their DNS client cache the previous hsot name and you won't be onlnie until the CAS servers notice the new IP address that they need to contact. In this case that you can not continue handling the same IP address is recomendable that you use a little TTL for the DNS record of your Mailbox Servers. Example TTL = 30 seconds

- If possible... try that the subnet used for the replication network be a different subnet in comparision to the Public segment where belong the mailbox servers. In this way you will avoid some possible design issues and more independence in the communication

Hello,

I do believe with AKHATER. The cluster i.e. DAG is intelligent to pick the replication network according to the NIC settings. There is no such manuall configuration required.

Regards,
:)

ok,, this is what i have for testing;

Server one Subnet A with two Nic
MAPI nic: 192.168.1.10 with Default Gateway
Replication nic: 192.168.1.20 without Default Gateway

Server two subnet B with two Nic:
MaPI Nic: 192.168.2.10 with Default Gateway
Replicaiton nic: 192.168.2.20 without Default Gateway

When i add Server one to DAG the The Failover Manager only shows the subnet and the MAPI Nic:
Subnets: 192.168.1.0/24
Network Interfaces: 192.168.1.10

Error message when try to add the second nice (replication) on EMC.

What should be my configuration with two nic in the same server?

@schnellsolutions although that what you are saying is true that is still not a reason to disable replication on the MAPI network unless there is a separate WAN/LAB for the network replication and this is done to save bandwidth on the non replication WAN.

@lena_mo do you have your static routes in place ? can the servers talk to each others on the replication NIC ?

I haven't confingure any static routes it?

and without static route how are you planning to make the 2 replication network communicate together ?

i haven't add the static routes but was plan to. But i my question is that, the failover cluster manager only shows MAPI nic not the other one. I can't add it to. i plan to use one nice for replication the other will configure as team network for MAPI.

If the network card interface is active, It should appear listered in the failover cluster manager. In which one of the network containners it is not listed?

It only can detects one network card if both are in the same subnet. It will detect both network cards with different network configuration. My original configuring was both of the nic are in the same subnet;192.168.1.0/24,, MAPI: 192.168.1.10, Replicaiton: 192.168.1.20

Mapiand rep,ication should not be in the same subnet

Yup,, that's what i was Missed

Thanks to All!!

I have a spareate question regarding of security of the OWA access which i might need to create a new ticket for.

Anyone know which will be the best option to setup OWA access for exchange 2010 in front of the firewall? Our current 2003 OWA was setup as a front end server in DMZ. I checked most the doc exchange 2010 doesn't not support in DMZ .

to be fair this is a totally different issue, i will have to ask you to open a new question for it