asked on

Exchange 2010 Hub Transport fail over???

Hello,

I am plannnig to move my exchange 2007 to a 2 server exchange 2010.

I will be doing a full install on both boxes.

I want to set up a CAS array and DAG between the 2 servers.
QUESTION..

Where does HUB transport fall into this? I need to be able to pull one server offline and have everything crank right along. How do I handle this with smpt send/receive to the net?

Thanks

Tony J

If you're using DAG's the best configuration is to have CAS and HUB on the same server.

So - two mailbox servers with DAG's (Don't forget you require Windows Enterprise for the clustering)

You then have two other servers. These servers have the CAS role installed (most likely in a CAS array) and the HUB role installed.

Brentologist

ASKER

2 servers total

both servers have all roles.

No edge transport servers.

How do I set up Hub transport for fault tolerance?

tigermatt

>> How do I set up Hub transport for fault tolerance?

Hub Transport servers will load balance their internal load themselves. If one falls over, another available one in the site will simply be found to handle mail submission internally.

For email being delivered externally, by far the easiest way is two separate public IP addresses linked to two public names (hub1.domain.com and hub2.domain.com). For your domains, set the MX records to be both hub1 and hub2.domain.com, either with equal priority (some basic load balancing) or one higher priority than the other, so it is used preferentially.

Bear in mind there are issues with building a DAG with Hub and CAS on only two Exchange Servers. You cannot load balance CAS traffic using NLB because this role cannot be installed with failover clustering, which will already be installed to support the DAG. There are a couple of options around this:

* Use a Hardware Load Balancer in front of the CAS boxes and point inbound CAS traffic at that. These aren't actually THAT expensive, if you're willing to move away from the bigger names. Have a look at http://w.e-e.com/np1Tfw for info.

* The "poor man's" solution is two entries for the DNS record of your CAS array in DNS. If you use outlook.domain.local as the CAS array name, you enter two DNS records in DNS, with a TTL of 5 minutes, one for the IP of each CAS server. If you lose a CAS you must manually remove the failed server's DNS entry and clients will reconnect to the remaining one soon thereafter. This is good for internal failover of Outlook sessions.

* A Forefront TMG farm sat in front of your CAS. This can do load balancing of HTTP traffic, but it doesn't solve internal MAPI access (via Outlook). For that, you'd need one of the above solutions, or I have seen all the Outlook clients forced into Outlook Anywhere internally, which then connects over HTTP through the TMG and has failover support.

Matt

Brentologist

ASKER

Is this possible:

http://imgur.com/BiZtx.jpg

ASKER CERTIFIED SOLUTION

tigermatt

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

Brentologist

ASKER

almmmmmmmmmmost done, i promise.

Thank you so far, btw.

So we got inbound smtp covered with 2 mx records.
internal failover covered with a short ttl and manual repoint on dns

What about outbound smtp? on the two boxes? how does that work here?

were done after this i promise

tigermatt

The questions aren't a problem. :)

For outbound email, create a Send Connector which has both your Hub Transport servers listed as Source Servers.

This means email can be sent through that send connector by either server and builds in the redundancy you need for outbound email.

Matt