News
Ian Masters, Double-Take Software: Replication to protect business-critical email systems
February 2008 by Ian Masters, UK sales and marketing director, Double-Take Software
E-mail, shared calendars and contacts are used to power some of the most crucial business-critical operations. While it was once just a convenient way for employees to communicate internally, today e-mail systems like Exchange are tightly integrated with other important applications and are one of the primary methods for communicating with suppliers and current or prospective customers. As a result, protecting email systems against costly downtime is a matter of huge importance for IT departments.
The value of email and messaging systems
The reasons to protect email messaging systems may seem self-evident but, to highlight the importance of system availability, the monetary implications are worth analysing. The value of any given data-set may be difficult to calculate, but the cost of a single outage can easily be determined. In most organisations, there are, at the very least, one or two subsets of end-users who are entirely dependent on these systems. Even if these groups do not produce revenue in the form of direct sales or billable engagements, salaries, benefits and fixed costs still accrue during an outage. The loss of the messaging systems for even a few hours will result in significant budget outlay without recouping any productivity. For revenue-generating groups the cost of this downtime is even more easily quantifiable. Avoiding just one outage is not only a good idea for the IT department, but for the CFO’s office as well.
Protecting Exchange
As of writing, there are several supported versions of Exchange in use and those versions can run on various configurations of the Windows® operating system. To each of these permutations can be added the complexity of service packs and individual hardware platforms. The result of these options is a wide variety of systems to be protected. The only common denominator is that the various Exchange files are stored on Microsoft Windows file systems.
An Exchange system is actually a complex configuration of multiple databases, used to store and manipulate data along with a message transport system to move e-mail and other information into and out of these databases. Protection of the database files, along with their log files and checkpoint files, is critical to the recovery of any Exchange server. Only by protecting all of these files can you be sure you can resurrect an Exchange system that has suffered a disaster.
To further complicate continuity, in many environments solutions such as e-mail archiving are deployed as an integral part of a company’s e-mail architecture. These tools are designed to remove attachments and outdated e-mails from the Exchange server in order to free up disk space and enhance overall performance. As a result it is possible that not all vital data actually resides on the Exchange server itself. The same applies to Blackberry and other mobile information systems that integrate with Microsoft Exchange to provide remote access to e-mail. Without a Disaster Recovery plan that accommodates these systems, restoring the Exchange server itself is only one step on the road to recovering the entire messaging system.
Restoring Exchange from Tape
When a recovery from tape is attempted, the tape systems will replace Exchange files onto the original server or a new server, at which point management tools can be invoked to re-constitute the databases to the state they were in when the tape backup was taken. This is an acceptable way to recover from something like a virus attack or human error where a point-in-time copy is needed to restore a good copy of the data.
However, for the majority of outages, like hardware failures and site-wide disasters, it is not the most effective means of restoring a failed server. Not only does it take a long time to restore from tape, messaging activity between the last tape backup cycle and completion of a successful restore is either lost of disabled. The extended Recovery Time Objective (RTO) and Recovery Point Objective (RPO) of restoring Exchange from tape is unacceptable to many businesses.
Data Replication for Exchange
While traditional tape backup is an excellent tool for long-term archiving, rapid recovery from an outage requires real-time data protection. Having such a solution, which is cost-effective, hardware independent and scalable, is something every IT manager should seriously consider.
One of the strengths of data replication technology is that it protects files at the byte-level. When Exchange writes data to any of its files, the actual byte-level changes it makes to the Windows file system are sent (replicated) to another Windows server. Once the data is protected on the second server, multiple options are available for achieving high availability and disaster recovery goals. The important point to consider is that it all starts with the data.
As with many database systems, Exchange requires that each transaction written to disk is performed in an explicit order. This is tracked continuously so that Exchange can maintain which changes have been requested and which have actually been committed to the physical database files. Generally referred to as a "transactional database system", these databases require replication systems that guarantee that write-order integrity of the data is maintained. Replication technology that utilises both transactional data integrity and write-order preservation processes allow the Exchange database system to immediately recognise a consistent copy of the data on recovery systems.
Protecting Exchange in your business
Exchange systems are available in a variety of flavours and configurations to meet nearly every business need and budget. Exchange can be run on stand-alone through to complex server cluster environments. Replication technology is available to protect each of those configurations. However, the easiest way to understand the concepts involved is to use a simple example.
Stand-Alone Exchange server (SAES) systems use a single server running all components of the Exchange system on the same physical or virtual system. This means the server acts as a mail-transport system, routing system, SMTP server and gateway and mail and public folder database server. Replication technology can provide data replication of all key information for the Exchange system and any other systems running on the production server and prepare a secondary server to take over in the event of a loss of the primary.
If the chosen replication solution includes failover functionality it will either automatically initiate a failover if the production server is unreachable, or alert the IT administrator to initiate the failover manually. Regardless of which failover methodology is chosen, the procedure for restoring services for end-users should be same.
First, the chosen solution will re-route end-users and dynamically re-assign all mailboxes from the failed server to the recovery server. This will allow end-users to regain access to their information the next time they attempt to connect to Exchange. This failover process is very fast and creates a relatively small load on IT resources.
Finally, the solution will prepare and start the appropriate Exchange and any third party services (anti-virus, mobile mail, archiving, etc) on the recovery server. This last step re-establishes a live Exchange server for end-users to connect to. Outlook clients may need to be re-started, but no end-user configuration will be required. In addition, other components such as Outlook Web Access (OWA) and other e-mail integrated system will simply pick up where they left off.
Email has become a mission critical application for most businesses in today’s world and being without it can affect business productivity, profits and reputation. When considering high availability or disaster recovery solutions for email resilience it can be difficult to find a technology that provides you with the required RPO and RTO and works with your existing infrastructure. Data replication may provide the answer, ensure data-on-demand and application resilience, don’t leave it to chance!
