What would you do if a critical server experienced a major hardware failure that resulted in a loss of data? How quickly could you restore the system from backup? What would be the loss to business while you restored the data? All IT departments typically back up critical servers, but IT professionals don't always give enough consideration to how long it would take to recover from a disaster.
As a network administrator at a healthcare facility that is facing HIPAA security regulations, I have had to think a lot more about these types of questions. Recently, I came across a product called Double-Take that could be very useful in a disaster recovery situation. Double-Take is a product of NSI software that allows selected files or an entire volume to be "replicated" to another server, effectively allowing a real-time copy of data to be maintained in case of disaster. Additional features such as failover capabilities allow even faster system recovery in the event of a failure. I'm going to show you how to install and set up Double-Take and how to use it to replicate data between two servers.
If you'd like to follow along with the steps in this article, you can get a 30-day fully functioning trial version of Double-Take from the NSI Software Web site. NSI also provides white papers on using Double-Take with Microsoft Exchange and SQL Server. Double-Take is designed to work with Windows 2000 and Windows Server 2003 systems.
Installation is straightforward. The setup provides options to Select Client And Server Components or Client Components only. The Client component option (which you'll probably want to install on your administrator's workstation) allows installation of the Double-Take management console, failover console, and a text-based client. The Client And Server Components option adds the Double-Take server to the install (this is the installation option that you'll run on the servers). An additional prompt is provided to select a hard drive location for the disk queue. The queue is a temporary location to store data that is being replicated.
Setup and management
Once installed, Double-Take functions are handled through the management console. The management console is an MMC console used to configure and manage Double-Take operations. Double-Take also includes a text-based client that can be used to manage Double-Take operation through a special scripting language. A failover console is also included in the install. We're going to focus on the replication aspect of Double-Take.
I chose to run the Double-Take management console from my workstation. The first time the console is launched from a workstation, Double-Take servers must be added. Right-click on the Double-Take servers and select New | Server. Type the new server name and click OK (Figure A).
If you are running the management console from a Double-Take server, the server will automatically be on the list. To manage a server you must log on. Right-click on the server and select Logon. Double-Take automatically creates local groups on the server that it is installed on. You must have administrator rights to manage a Double-Take server. After logging on, you can further drill down and see the resources available on each server (Figure B).
Now let's set up replication between two servers. Right-click on one of the servers and select New | Replication Set. A New Replication Set icon is added below the server. You can name the replication set anything you choose. Drill down as needed and select the drives and/or folders you wish to replicate (Figure C).
Next, right-click on the new replication set and select Connection Manager. The Connection Manager tool (Figure D) is used to set up a connection between two servers.
The first tab on the Connection Manager screen is the Servers tab. This allows selection of the source and destination machines. A server can have an all-to-one replication relationship, meaning all replication is sent to a specific location on the destination server, or a one-to-one, where all replication is sent to the same location on the destination server as it resides on the source server.
Double-Take begins the replication process by mirroring the replication data from the source to the target. Mirroring assures that the replicated data is an exact copy of the source data. At the completion of mirroring, replication takes over and keeps the two copies in sync.
The Mirroring tab allows fine-tuning of the mirroring process (Figure E). The file difference with checksum allows the file checksums to be compared to assure the files are the same. This can be helpful when replicating a database file, since the file size and date stamp may not change as new data is added. Comparing the checksum assures that you have the latest changes.
During the replication process, it is possible that some files may no longer exist on the source but were previously copied to the target. This creates an orphan file. Double-Take allows logging and saving of orphan files for review. The Orphans tab (Figure F) controls whether orphans are saved or deleted.
One concern with replication is how much traffic it will generate and what impact it will have on the network. The Transmit tab (Figure G) allows control over how much bandwidth is used based on the speed of your network backbone, and it can also set a specific start and stop time to be defined. In my testing, I used a setting of 50 percent over a 100-Mbps backbone with little noticeable impact to my users.
The Verify tab (Figure H) allows a verification interval to be set along with a setting for how often the data is reverified. Verification is optional. As with the mirror tab, the best practice is to use the block checksum option to assure the source and target files are identical.
The last tab is Failover. This allows setting Double-Take to perform a failover operation in the event that the source server goes down. Keep in mind that failover is optional. Replication can be configured without enabling the failover option.
After all the options are set, click Connect and the replication process will begin. The Double-Take console will now display the status of the connection (Figure I). When the mirror status reaches 100%, replication takes over and keeps the data updated.
Other uses for Double-Take
While evaluating Double-Take, a colleague and I were in the planning stages of a server migration. We were migrating our HIS (hospital information system) from old Compaq hardware to new Dell servers on a SAN. The migration required moving database files from the old servers to the new ones.
In the past, we had planned system downtime of several hours while we copied the data over the network. It looked like this time around we were going to have to inform our users that we would have to take the system down for several hours at a time over a period of several days. And then suddenly it hit us: To minimize the downtime, why not try using Double-Take to mirror the data in real time while the users were still using the system? The result was a maximum of 15 minutes of downtime for each server while we verified the database sizes and renamed the servers. This greatly reduced the expected downtime and made the users and management very happy.
HIPAA goals accomplished
What prompted our evaluation of Double-Take was our planning for a disaster recovery solution. The HIPAA security regulations require that we have a disaster recovery plan and that we test the plan. We have an off-site facility that is connected via a 45-Mbps wireless link. What we wanted to accomplish was an off-site version of our hospital information system.
The first challenge was space. Our current hospital information system is installed on 13 servers connected to a SAN. We did not have the resources or the space to house an identical system off-site. To duplicate the servers off-site, we decided on three blade servers, with each blade server accommodating six servers. The blade servers take only nine rack units compared to a full-size 42U rack to house the current servers and SAN hardware. The leftover blade servers will provide an off-site domain controller, e-mail, and other business applications. Double-Take will be installed on each machine, replicating the data over the 45-Mbps link. In the event of a disaster, we can use Double-Take to restore the failed servers and data or use the failover option to allow an off-site server to stand in for a failed one.
I found Double-Take to be extremely easy to set up and use. The download also includes a well-written manual, although absent from the manual were instructions on installing the software. The installation was easy, but I usually expect to find that in the included documentation.
The list price of Double-Take is $2,500 per copy, and it takes two copies for a working system. Double-Take is sold through representative companies throughout the U.S. and abroad. Sun-Belt software provided my copy and handled my tech support questions. I found them to be quite knowledgeable about the product. If you are in charge of critical servers and need a solid disaster recovery plan, I recommend giving Double-Take a look.