So after a long silent period apart from Outlook live lab videos on the Exchange blog, suddenly there is Exchange 2010 beta!

So what can you expect in the new industry standard mail platform, this is part 1.

Well the major change is in High Availability, take LCR, CCR and SCR
from Exchange 2007, mix it all together and you have the best of all
three! So there is no more LCR, CCR and SCR in Exchange 2010, but it is
called Database Availability Group and so yes another acronym: DAG.

The major change is that Windows Failover Clustering is not involved
in DAG, the high availability is at database level. You can have up to
16 (!) copies of each mailbox database spread over you mailbox server
farm. There are no more storage groups, you now set the databases at
organization level and not at server level.

Well that sounds really enterprise/datacenter level, well with just
2 servers you can have high availability! without the need of expensive
cluster storage solutions. Copies of databases may be placed on servers
that have other exchange servers installed.

Another major improvement that benefits DAG is the 50-70%
performance gain in IOPS. So there is no need for expensive fast
spindles for storage, you can use direct storage, even desktop class
SATA disks. So with this cheap storage you can give your users larger
mailboxes. Microsoft even claims you can use the cheaper SATA drives
without RAID, since you have copies of the databases on other servers.
So now you can have a performing high availability exchange solution at
home 🙂

But of course DAG works best in a datacenter environment, with 16
copies per database you can have site redundancy spread all over the
world.

There has been no announcement yet about the Exchange SKUs, so I
don’t know what form of DAG will be available in the standard edition,
but LCR was available in Exchange 2007 in the standard edition which
requires a manual recovery. I assume that 2010 standard edition will
allow at least 1 database copy with automatic recovery support.

Another part of high availability is the new move-mailbox feature:
online mailbox move. Now the end user can continue working with their
email, reading, sending and receiving, while the administrator moves
the mailbox to another server, all just during work hours. At the last
stage the end user will expect a interruption, when the last sent and
received email is copied over to the new location. So this is like
Vmotion on VMware ESX (and soon Live Migration on Hyper-V

Functional Descriptions

Database Availability Group: A set of Mailbox
servers that uses continuous replication to provide automatic recovery
from a variety of failures (disk level, server level, datacenter level).

Database-Level Failover: Exchange Server Database
Availability Groups provide automatic failover at the database level,
without the complexity of traditional clustering. A database-level
disruption, such as a disk failure, no longer affects all the users on
a server. Because there is no longer a strong tie between databases and
servers, it is easy to move between database copies as disks fail. This
change, coupled with faster failover times (30 seconds), dramatically
improves an organization’s overall uptime.

Improved Site Resiliency: Exchange Server Database
Availability Groups makes it easier to implement site resilience by
simplifying the process to extend data replication between datacenters
to achieve site failover. Log files are also compressed to improve
transmission time and reduce network bandwidth usage.

Easier Deployment: Administrators can add high
availability to their Exchange environment after their initial
deployment, without reinstalling servers. Small organizations can
deploy a simple two-server configuration that provides full redundancy
of mailbox data along with Client Access and Hub Transport roles. These
changes put high availability within the reach of organizations that
once considered it impractical.

Integrated Cluster Administration: Exchange Server
Database Availability Groups feature automatic failover without the
complexity of traditional clustering. The proven capabilities of
Windows clustering are integrated with Exchange and are transparent to
the administrator. Administrators no longer need to master clustering
concepts or deal with separate administration tools in order to provide
enterprise-class uptime.

Backup-less Support: The Exchange Server Database
Availability Group architecture allows log file replay to be lagged,
enabling administrators to perform point-in-time database restores
without the need for tapes. Organizations can rely on their high
availability infrastructure rather than tape backups to recover from
failures, and substantially decrease their operating costs.

Transport Resiliency: Transport servers in Exchange
Server 2010 feature built-in protection against the loss of message
queues due to disk or server failure. Servers retain a "shadow" copy of
each mail item after it is delivered to the next hop inside the
organization. If the subsequent hop fails before reporting successful
delivery, the message is resubmitted through a different route.

Advertisements