What you will learn from this tip: Most organizations back up data one way or another nowadays -- but just how many do it well? This tip will show you some of the areas of backup you should consider optimizing.
There is definitely a line to be drawn between data backups and good data backups. If your organization does not yet back up data and you just stumbled on to this storage site by mistake, the moral of the story is simple: Start backing up now! For the rest of us, there are a number of items that are often overlooked when performing data backups. Many administrators focus on backup success or failure and performance but often forget about the most important aspect of backups: restorability.
Optimizing backups is not always about storage device performance and tweaking software configuration settings. Here are some other aspects to consider:
Know what you are backing up: Are you backing up everything you should be? Do you back up more than you need? There might be critical workstations or PCs that are missed daily while users assume they are covered. Conversely, you might be consistently backing up static or redundant data (i.e., some OS files, archived database tables or exports). A series of discussions with the business side may provide some valuable insight.
Disk backup: D2D backups or data replication are probably some of the best ways to optimize backups (and restores) from a performance perspective. Unfortunately, bandwidth for remote backups and long-term storage capacity costs remain major hurdles for most organizations. This capability should be deployed based on data and application criticality.
Prioritization: Data must be categorized based on application restore priority. The application criticality dictates the RTO and drives the priority ranking. This same RTO and restore priority should be used when choosing the backup methodology (i.e., D2D backups or replication). This warrants further discussion with the business.
Full and incremental backups: Many backup products are now "incremental always" capable. Some of these products also offer the ability to create "synthetic full backups" by concatenating incremental backups into a current state image. This feature can dramatically reduce the amount of storage as well as time spent backing up unchanged files.
Number of tape drives: When backing up to tape, the number of tape drives should be equivalent to the number of desired (or required) concurrent data streams. However, network bandwidth must also be considered when adding tape devices.
Network and system performance: Ensure that the network bandwidth and disk subsystem will be capable of handling an unusually large amount of data, such as simultaneous full restore streams. While full backups are typically scattered throughout a week, simultaneous full restores may not offer the same flexibility. LAN-free backups (SAN) can also offer an alternative backup and restore path.
Data interleaving: Avoid using interleaving to optimize backup performance. It may be tempting as a means to increase the number of simultaneous backup and restore streams, but it carries a cost in terms of performance.
Scheduling: Backup and administrative task schedules should be reviewed for opportunities to spread the workload over a 24-hour cycle when possible.
Monitoring: Proactive monitoring is by far the best way to ensure and maintain backup and restore optimization. Backup products all offer various levels of monitoring capabilities that can be enhanced with third-party products.
For more information:
Optimizing tape restores
About the author: Pierre Dorion is a certified business continuity professional for Mainland Information Systems Inc.