By Mike Talon
week's interactive scenario described a file server high availability (HA) solution with a fatal flaw. As of Thursday, May 29, no one had addressed the one specific issue that would absolutely cause the scenario to fail during a failover: There was no redundant domain controller—either a backup domain controller (BDC) for Windows NT or an Active Directory connector (ADC) for Windows 2000.
Without a redundant domain controller, if a failure took out the main data center, the company would experience quite a few problems, such as security and access difficulties. This actually happened to a client of mine in a Windows 2000 network, and it wasn't a pretty sight to see.
When looking at HA for file servers, you might think they're the easiest solutions to configure. After all, applications don't run directly on file servers. How difficult could it be to set up and maintain the DR and HA systems?
Although scripting and failover time systems are lessened with file-only servers, the systems that need to be set up ahead of time—preferably when the solution itself is configured—can be more extensive than you bargained for. Primarily, you'll need to make sure you have systems in place to handle file shares, security limitations, and routing.
File shares are the heart of every file server. Without them, end users will have a hard time locating their files after failover. There are a few ways to make sure your file shares cross over to the other side of the DR solution. The easiest way is to set up the shares ahead of time on the DR server. Since that server has a different name, clients won't route to this machine until failover is enacted.
Once a failover scenario is met, you can rename the DR server, change login scripts to remap user drives to the new server, and/or utilize a software package that's designed to connect users to their shares on the new system. Ask your DR vendors about software packages that handle failover and remapping as part of the DR system. These systems eliminate the need to set up the file shares because the software takes care of it for you.
Security limitations are a major concern when discussing HA and DR solutions for file servers. Once the remote file server comes online, make sure that everyone has access to the correct files they need to do their jobs and are prohibited from the rest of the file server.
Verify that the domain structures—and other security formats—exist on both sides of the DR solution: production and backup. This is accomplished by stretching the security borders between sites with VPN and other encryption solutions or (in the Windows world) establishing domain controllers in multiple locations.
This will ensure that all the accounts necessary for proper file access are available when disaster hits and the solutions fail over. You can also manually create the security accounts on both sides of the solution, but it's very time-consuming and doesn't always work, depending on the topology of the network and the design of the security infrastructure.
Once failover occurs, route your clients to the appropriate servers. Either offer multiple connections for end users to choose from, or allow for multiple VPN tunnels for offices to utilize. Both of these solutions require that your staff is ready to jump in and assist, since most of your end users won't be able to handle this part of the failover operation. This procedure may take some time to complete, so keep that in mind when drafting your SLA regarding recovery time.
In addition to the usual DR and HA preparations, it's important to take these special considerations into account. With your systems fully prepared, you can rest assured that failing over the file servers won't cause a failure all their own.
Mike Talon is an IT consultant and freelance journalist who has worked for both traditional businesses and dot-com startups.