Product Reviews

Still Beating

Neverfail can keep your Exchange server running through just about any disaster.

• By Ben Brady
• 05/01/2006

Five years after living through a particularly nasty e-mail outage due to a drive failure, I'm more dependent on e-mail than ever. That's true of just about anyone. In fact, most companies would be lost without their e-mail, the primary means of communication between customers, employees, partners, suppliers and vendors. And Microsoft Exchange servers are critical for maintaining those open lines of communication.

Neverfail Heartbeat is designed to keep those servers running -- no matter what. It's a high availability package for critical Microsoft server applications such as Exchange, SQL Server, IIS, SharePoint and file servers. There's also a module to support BlackBerry Enterprise servers.

Neverfail uses a cloned server approach to achieve high availability. In a typical Neverfail setup, your primary server hosts your critical data and applications and a secondary server functions as your failover unit (with the "cloned" data).

Before starting, you should run Neverfail's Server Check Optimization Performance Evaluation (SCOPE) on both your servers. This inspects your hardware, services, applications, network connectivity, utilization and system performance. Gathering this data helps ensure there are no performance issues or conflicts. It also helps establish a baseline for later performance monitoring.

After analyzing your SCOPE data, you'll be ready to install. The documentation is very clear and concise. I was able to quickly install the package on my primary server. I encountered a small hurdle installing on the secondary server, as I was using different types of hardware.

My testing scenario included Exchange and all the transient working files, Registry entries, data stores and so on. Using Windows Backup during the Neverfail installation on the primary server, Neverfail creates a backup copy of all the related system files, applications, data files and Registry entries for keeping the Exchange server running.

After backing up the Exchange server data, Neverfail installs on the secondary server where the backup is restored, creating the cloned server. After completing installation and connecting a channel between the two servers, it replicates data from the primary server to the secondary server to keep the "clone" current and intact. One safeguard you might want to consider here is Neverfail's Data Rollback module, which can help you recover from replicating bad data.

[Click on image for larger view.]

Figure 1. You'll do most of your day-to-day administration tasks in Neverfail Heartbeat's Management Client GUI.

Total Control
The Neverfail client GUI is where you'll do most of the day-to-day administration. You can use this from either server or from any workstation within the same subnet as the primary server.

From the client GUI, you can control the activities and behavior of your servers. The servers will each assume a role of either "active" or "passive" server. Right after installation, your primary server will be your active server, and your secondary server will be your passive server.

You can maintain and monitor the connection between the servers, the status of the replication, whether or not the data is currently synchronized, start and shut down the service and manually start a switchover from your active to passive server. The interface also lets you maintain logging information, view and parse logs, e-mail logs to admins and set up alerts based on certain application or performance attributes.

[Click on image for larger view.]

Figure 2. Deeper within the Management Client GUI, you can configure the connection between your primary and secondary servers.

The GUI also lets you monitor other applications. For example, in the event of a failover to your secondary server, you probably wouldn't want your Exchange server up and running without your anti-virus and anti-spam applications running.

You can configure Neverfail to monitor services, performance and network connectivity. You can also control the various tolerances of these attributes and how Neverfail reacts to these triggers.

Outage Time
Neverfail's primary purpose is to ensure high availability of Exchange servers. To fully test this claim, I yanked out the power cords. The box immediately went dead. The clients in my lab all had Outlook open and were connected to the Exchange server when it went down. Each client was set to check mail every minute.

The failover to the secondary passive server took less than two minutes. Going around to each of the clients, I was able to send and receive, update tasks, make meeting requests and calendar updates.

That was easy enough, but it was only half of the process. After restoring power to my primary server, I had to restart the Neverfail service. From the client GUI, I initiated a "switch back" and monitored its progress as the secondary server prepared for the switch and made the change. Once again, my Outlook clients lost connection for a little more than a minute, then everything was back up and running on my primary server.

[Click on image for larger view.]

Figure 3. Neverfail Heartbeat’s straightforward wizard guides you through setting up your servers.

Because a power failure isn't the only thing that can happen to a server, I tried a few more tests. I turned off a couple of Exchange services. After about a minute, they were all back on again. Then I unplugged the network cable; again, all my Outlook clients were back up and running after about one minute.

Switching back to my primary server after network connectivity was restored was as seamless as before. I received several alert messages on the secondary server each time it detected a failover trigger. You can configure the outage thresholds and the response time for initiating failovers.

I tried several other methods of taking the Exchange server offline, including a manual "switch over" from the client GUI without any problems. Finally, I decided to stop all Exchange services on the primary server, disconnect both the network cable and the channel cable between the two servers and then quickly shut off the power to both the primary and secondary servers.

Later, I reconnected all the cables and powered and booted the machines. Everything started up as expected.

I found just a few small quirks during my testing. Both the primary and secondary servers use the same IP address, but Neverfail installs a packet filter on the passive server to avoid IP address conflicts. This may interfere with some types of traditional troubleshooting, so keep that in mind.

Also, some of the services you would expect to be running on the passive server will not run under normal operations. This may irritate some administrators who prefer to start those services manually or set them to "Automatic."

A Healthy Heart
I had a very positive experience with Neverfail. Regardless of the various types of outages I threw at it, it helped my Exchange servers bounce right back.