Exchange Server 2007 Log Shipping through Heartbeat Network
Imagine a scenario where we would have an Exchange Server 2007 CCR on Windows Server 2008. In this setup we only have two network cards.
Initially our thoughts would go to that it’s a best practice to configure the private/heartbeat NIC as the primary one to be used for log shipping and seeding and thus to move that traffic away from the public/MAPI NIC, as per Tim McMichael’s blog.
So the main question here would be if we could use such configuration. Unfortunately there's no support statement on this, just best practices. But we can't configure log traffic on a network that doesn't support client traffic, so a purely private network would be restricted from log shipping anyway.
As per the above blog instructions the heartbeat NIC needs to be set to mixed as well which bring us to our initial question again:
Exchange 2007 SP1 CCR (Cluster Continuous Replication) – Two Network Ports
When using a cluster continuous replication cluster with two network ports, your options are limited. Consider the following design:
First port set to “allow the cluster to use this network” and “allow clients to connect through this network”.
Second port set to “allow the cluster to use this network” and “allow clients to connect through this network”.
You’ll noticed that in this configuration both networks are set to “allow clients to connect through this network”. This is necessary in order to establish the private network for use with log shipping functions.
To establish this network for log shipping functions, refer to the enable-continuousreplicationhostnames cmdlet:
If used the replication service will first prefer to perform log shipping functions over the “private” network. Should the private network be unavailable the replication service will resume log shipping functions over the public network.
However in the above case, if we didn't enable both cluster and client connectivity for both networks, we'd have at least one single point of failure again. That's something to avoid in a two-port configuration. I believe Tim McMichael’s blog generally recommends a minimum of three ports for CCR systems, with two networks set for public and private and one for private only.
Last but definitely not least, for Windows Server 2008 clusters there really is no such thing as a dedicated private network anymore. As soon as you enable the network for client use, you have to enable it for cluster use, you can’t enable it for client use until you have set it for cluster use. Additionally there is no way to control the priority of the networks for cluster use, the failover cluster adapter determines the routing paths for that. Also, apparently “the allow clients to connect through this network” option is only used for automating and/or controlling the wizards for configuring services and applications, this setting tells the wizards which networks to create the service IPs on. So it actually no longer matters if you don’t have a dedicated heartbeat network, even ExBPA has been updated to not fire the “Dedicated heartbeat network not found” warning for Windows 2008 Failover Clusters. The key thing is that the networks must be on different subnets, and therefore client traffic would never go over the cluster/log shipping network.
The main reason we wanted dedicated private networks in Windows Server 2003 was because we were not very tolerant of missed heartbeats, in Windows Server 2008 we are certainly far more tolerant of missed heartbeats with the new network heartbeat improvements, especially when using a Majority Node Set with File Share Witness (no bus resets to rely on, with buggy disk drivers and firmware). But if you are wanting to be extra, extra safe then go with three networks.