By Chris Lehr, Senior Microsoft Consultant
At Simpler-Webb, I've deployed about 30 Exchange 2007 organizations since its release. Of those, under 10% of them addressed high availability or disaster recovery planning. Here's why and what 2010 does to address those needs.
Exchange 2007
First, in Exchange 2007, if you wanted a highly available solution, you needed to design it as such from the beginning. If you deployed a single server with all three mailbox roles, you were not able to easily add a second or third server. You needed to configure a new cluster and move mailboxes to it.
Second, when talking High Availability in Exchange 2007, you were talking CCR, or continuous cluster replication. Unfortunately, when designing a CCR solution, only the mailbox role could be clustered. The Hub Transport and Client Access Server roles needed to be deployed elsewhere. The obvious choice for many was an additional piece of hardware.
Even with a highly available mailbox role, the Hub Transport and Client Access Role still had a single point of failure. For most, the way to address this was to get a pair of Hub Transport/CAS servers and load balance them.
So, now we have a highly available four server Exchange organization, but still no Disaster Recovery. In Exchange 2007, this was brought in by SCR (Standby Continuous Replication) in service pack one. SCR allows for a server with an "offline" copy of your databases. Should a disaster be declared, this offline copy could be brought online and users would be able to access email again.
The major downsides of SCR were that it required a fair amount of technical Exchange knowledge to configure, and more importantly, to recover. When you are in the middle of a disaster, this is not the time to type a lot of commands with shaking fingers!
Exchange 2010
Exchange 2010 essentially combined the best of CCR and SCR into one technology, known now as Database Availability Groups (DAGs).
The two biggest improvements of Exchange 2010 DAGs over Exchange 2007 CCR/SCR are that we can now do an incremental deployment and we do not need to split the Hub Transport and Client Access roles off.
This allows for a single server Exchange 2010 environment today with the ability to add a second server next year and later add a third. You can go up to as many 16 copies of your database. Each additional copy increases your high availability possibilities. Additionally, leaving the HT and CAS roles on the same servers means that your licensing and server costs are lower than they would be in an Exchange 2007 environment.
In addition, this same DAG technology can also include DR planning. A single server at a DR location with a copy of your databases can be brought online in the case of a disaster.
As you can see, Exchange 2010 allows you to deploy both a highly available and fault tolerant Exchange environment in fewer steps, with easier administration. In addition, Exchange 2010 has a 70% reduction in IOPS in storage requirements, so the mailbox servers you are utilizing can utilize cheaper storage options. In addition, with three or more copies of your database, Microsoft even recommends moving away from RAID 5 or RAID 10 for your database disks, but instead model a single SATA disk per database and transaction log files.
Exchange 2010 has a lot to offer beyond high availability and disaster recovery. Also, we are currently scheduling both Exchange 2010 design plan projects and implementations and are excited to help our clients realize all the benefits of 2010. If you would like more information, please
contact us. Also, check out our
Exchange 2010 HA & DR video.