The three steps in building a tape backup
The tragic events of September 11 have made us all more aware of the devastating potential of a physical disaster. Many companies are looking carefully at their backup strategies to ensure data integrity, reliability, and business continuity in the event of a disaster.
Developing a good backup strategy requires thoughtful planning, from the selection of devices to the determination of backup schema and other logistical issues. Each step must be carefully analyzed in light of such factors as staffing, technical expertise, and budget. We’ll take a look at these factors and break down the requirements of a successful backup strategy.
Selecting your backup system
First, an organization must determine what backup platform (hardware) it will use for data protection. Many choices exist, from extensive data-replication systems that create off-site replicas to traditional tape-based systems. Budget and expertise will determine which method (or combination of methods) is best for your organization.
Tape systems are generally the least expensive and least technologically complex of the choices on the market today. Since tape systems will meet most of the audit and reporting requirements facing the modern enterprise, we’ll focus primarily on this methodology.
Before you can determine your backup strategy, you must first determine how much data your organization will be protecting. This information will help determine what type of system you require—from single-tape drives through jukebox or silo systems that can manage hundreds of tapes at a time.
The process of determining what will be protected goes beyond summing up the total of all data on your servers, since much of this data is likely not mission-critical to your organization. Prime examples are program and application files—which are easily restored from the original media—and users’ personal files, such as MP3s.
Once you know the amount of data you need to protect, you can determine which type of device best suits your needs. DDS tape systems are generally acceptable only for smaller organizations with less than about 10 GB of data to protect. For the small/home office, this is a very good choice, since the drives are economical and are considered reliable. Many jukebox or silo options are also available for this platform, allowing these venerable tapes to scale to meet most small to midsize business needs as well.
For the larger business, DLT tape systems are somewhat more expensive but have significantly greater capacity for data. DLT systems can be scaled to meet the data-protection needs of most organizations, regardless of size. Jukebox and silo systems are also available for this platform, including robotic and other forms of automated rotation systems to handle the needs of even the largest organizations.
Once you’ve decided on hardware, it’s time to set your sights on software. There are many excellent backup software systems on the market today. Each offers similar functionality, so take a look at interfaces, pricing, and other factors to determine which is right for your organization. Two of the leading products are ARCserve and BackupExec.
Setting up your methodology
Your next step is to determine the best methodology for backing up your data. Most commercial backup software systems allow for a wide array of options. The most common are the following:
- Full: This methodology transfers a copy of all data within the scope of the backup to tape, regardless of whether the data has been changed since the last backup was performed.
- Differential: This methodology backs up all files changed since the last full backup, regardless of whether they have been changed since the last backup operation of any kind.
- Incremental: Here, only those files that have changed since the last backup operation of any kind (full, differential, or incremental) will be transferred to tape.
Essentially, the method you choose will depend on your tape capacity and your need for a very speedy recovery. For example, by running a full backup daily, you will need a very large amount of tape capacity but will need only the last backup tape to restore all data.
Conversely, using a full backup once a week with an incremental daily lets you use much less tape space, but a recovery would require the last full backup and each incremental to be fed back onto the servers. Most organizations will utilize a combination of methodologies that allows them to conserve tape space while still allowing data to be quickly restored. Using weekly full backups with daily incremental backups is generally considered the strategic norm.
Rotating your tapes
Next you will need to develop a tape rotation scheme and decide where to store your tape media. Unfortunately, most companies store physical backup media within about three feet of the servers they protect. As recent physical disasters have shown us, this is not a prime way to protect data since the tapes stand a very good chance of being lost with the servers in a disaster. Physical media needs to be stored off-site in a secure location from which it can be retrieved quickly if needed.
To reduce costs for both media and off-site storage, a good tape-rotation scheme is essential. By reusing tapes after a predetermined period of time, tape rotation schemes ensure that a minimum number of tapes are stored off-site. Rotation schemes come in several flavors and complexities.
Perhaps the most common rotation scheme is Grandfather-Father-Son (GFS), which has been in use for quite some time and has proven to be effective. In the generic version of this scheme, full backups are done at two intervals. The first interval is monthly, and that media is immediately stored off-site (grandfather). Weekly full backups are also performed (father), and these are generally held on-site for the week of their use, then moved off-site for a predetermined number of weekly cycles (two or three cycles is generally sufficient).
Daily incremental backups are performed (son) and stored on-site for the week of their use as well, then moved off-site with the corresponding weekly backup on the same cycle schedule. At the end of a cycle for a set of tapes, those tapes can be returned to the system and reused. Each monthly tape would be held for 12 months before being reused.
Weekly tape sets are generally held for about three weeks after they leave the server room (so about four weeks total) and are then reused. The number of times a tape can be reused is determined by the type of media, so check with the manufacturer to ensure you don’t overuse a tape.
GFS rotation schemes allow for the monthly backups to be immediately stored in a secure location, while the most current weekly full backups and incrementals are housed for immediate use in restoration, then allowed to sit off-site in case of a catastrophic failure. While there is some limited liability inherent in this system (you could lose up to one week of data if you suffer a fire or some other disaster), this system offers the highest security with the lowest cost.
Of course, your organization may require more security, in which case you can always courier tapes off-site as they are generated. While this is a great system for data protection, keep in mind you will need to get those tapes back on-site for restoration purposes, which will increase courier and other costs.
By determining what data needs to be protected, what media to use for that protection, and how you will protect the media itself, you can create and maintain a reliable backup system for your organization. Such a backup system will ensure business continuity in the event of a disaster.
How do you plan to modify your current backup system?
We look forward to getting your input and hearing your experiences regarding this topic. Post a comment or a question about this article.