Download the authoritative guide: Enterprise Data Storage 2018: Optimizing Your Storage Infrastructure
In the first article in this two-part Storage Basic series, we looked at some of the general backup procedures used to manage the process of backing up large amounts of data. In this article we dig a little deeper into backup strategies, specifically looking at LAN-free and server-free backups and how they are used to eliminate the backup window. First, though, we’ll review two types of network designs and how they impact backup traffic.
When designing a network, careful consideration must be given to the location of backup devices, as their physical placement on a network has both performance and security implications. In a traditional network design, a backup device is located on the same segment as all of the other network devices.
As you might imagine, there are both advantages and disadvantages to this approach. When network devices are on the same segment as the backup device, it’s relatively easy to configure all network devices, including workstations, to back up to the backup device. It’s also a low-cost solution, as backup devices can be added to a segment without major changes to the network infrastructure.
However, because the backup device is on the same segment, there’s a greater security risk, as workstations may be able to directly access the backup server. Furthermore, if backup devices are located on the same segment, the volume of backup data being transferred can negatively impact the speed of the network. This is why backups are often performed in the off hours to prevent impacting network users. As discussed in the first part of this backup basics series, this creates the problem of the “backup window.”
Removing Backup from the LAN
For those organizations that transfer huge amounts of backup traffic, there is another option — physically remove the backup device from the LAN. In this scenario, the backup device is located on a dedicated Fibre Channel segment. This network design is typically more secure, as access from Ethernet network devices can be limited and controlled by the storage or network administrator.
Perhaps the greatest benefit of removing the backup device from the production LAN is in performance. LAN bandwidth is preserved for the production network, and backups can be performed at any time without impacting users.
On the downside, when backup traffic is taken off the regular LAN and placed on a high-speed alternative, it can be costly in terms of both dollars and administrative overhead. However, when in place, it provides the ability to provide two unique backup strategies: a LAN-free backup and a server-free backup.
In early storage development, a common backup strategy was to have backups performed to locally attached tape devices. For larger network environments, this approach had two significant drawbacks — cost and management. Maintaining a large number of decentralized tape drives and media became difficult and inefficient. This lead to the development of software that allowed designated servers to act as centralized storage devices. While this approach allowed backups to be managed from a central location, it also meant that all backup traffic had to travel over the LAN to the backup server.
In many environments, this can still be an effective backup design as long as the LAN can support the amount of traffic that the backups generate. However, in many of today’s modern environments, the amount of backup traffic that must cross the LAN is simply too great for the LAN regardless of the LAN technology used. The solution in this case is to take the data off of the LAN.