There are two main goals that should be kept in mind when planning a storage area network backup. The first of these goals is the elimination of a backup window. Modern data centers need to have data and applications accessible 24 hours a day, so a backup window is generally unacceptable. A second goal should be to reduce the impact of the backup on your network infrastructure as much as possible. After all, you don’t want to add an additional workload to a server or network resource that may be already overworked.
There are many different topologies that can help you to achieve your backup goals, and I will be discussing four of the most common approaches.
Image copies can best be thought of as mirrored disks. A SAN is able to create an image copies by mirroring a production disk to another disk inside the storage array. This copy process happens at the hardware level, which means that network servers are not involved in the image creation process. Of course because image copies are a hardware-level feature, they are not supported by all storage area networks.
Image copies alone are not typically a sufficient backup solution. The reason for this is that, if a storage array failure occurred, the failure could potentially affect both the production disk and the image copy disk. As such, image copies are only one part of the overall backup solution. Traditional backup software is usually used to make backups of the image copy disk. That way, backups can be made without impacting the performance of the production disk.
Another option for backing up a storage area network is to perform disk-to-disk backups. A disk-to-disk backup is simply a backup that is written to a disk or disk array instead of to a tape. In the case of a SAN, disk-to-disk backups usually treat a storage array as a virtual tape library. Server disks almost always outperform tape, so using a virtual tape library essentially allows data to be backed up much more quickly than it could be on a physical tape library.
As is the case with image copies, disk-to-disk backups are not the perfect solution because they are prone to storage array failures. As such, disk-based backups are typically written to another storage medium. For instance, many organizations use what is known as disk-to-disk-to-tape. This architecture involves periodically copying the contents of the virtual tape library to a physical tape that can be stored offsite. A new architecture called disk-to-disk-to-cloud is also starting to become popular. This architecture copies the contents of the virtual tape library to cloud storage.
Another option for backing up a storage area network is to use what is known as a serverless backup. As the name implies, a serverless backup is a backup in which the data does not have to pass through a server.
There are a couple of different variations on this particular backup architecture, but one of the more common methods involves using a protocol called E-copy. The E-copy protocol is an extension to the SCSI protocol. As such, using E-copy requires your SAN hardware and your backup software to be E-copy-aware.
The basic concept behind an E-copy based backup is that your backup software directs the data source to copy data through the SAN to the target device (usually a tape library). In doing so, the data is sent directly from the source to the destination. It does not pass through the backup server in the process. This helps to make the backup process more efficient.
One last SAN backup option that I want to talk about is a LAN-less backup. A LAN-less backup is a backup in which backup traffic is isolated to the SAN rather than passing through the corporate network. This differs from a serverless backup in that a serverless backup would not involve the application servers. In a LAN-less backup, the application servers contain backup agents and are involved in the backup process.
There are a couple of advantages to performing LAN-less backups. One advantage is that the LAN will not have to accommodate backup traffic. The other advantage is that data can pass through a SAN much more quickly than it can pass through a LAN because of the overhead involved in TCP/IP.
In order to do a LAN-less backup, you will need backup software that is SAN-aware. Most of the major vendors offer some form of SAN support. You will also need a tape library (or other backup target) that is directly connected to your SAN. Finally, your backup server must be connected to the SAN. The backup server’s job is to communicate with the other SAN-connected servers and orchestrate the backup process. In doing so, the entire backup process is isolated to the SAN. This should allow data to be streamed to your backup target more quickly than it could if your backups were being performed over the LAN.
About the author: Brien M. Posey, MCSE, has previously received Microsoft's MVP award for Exchange Server, Windows Server and Internet Information Server (IIS). Brien has served as CIO for a nationwide chain of hospitals and has been responsible for the Department of Information Management at Fort Knox. You can visit Brien's personal website at www.brienposey.com.
This was first published in May 2012