How to backup a networked fileserver?

Question

For work I'm trying to back up a networked file server. It's running Windows Server and serving all the files in our office. We have external drives to back it up to we just need some software to automate it for us. What kind of general recommendations can the SF crowd make as to what piece of software we should use and best practices for backing up a company networked file server?

Answer 1

General statements re: backup

I'm not going to get into specifics of backup technologies, software, operating systems, etc. in this part. Nor am I going to get into specific techniques and policies (various rotation schedules, using synthetic backup, etc). I'm going to stick to basics here, but I'd advise you, once you understand the basics, to dig deeper into how the requirements of your environment affect your needs.

Above all else, the point of backup is to be able to restore data. We don't run backups for backup's sake. Above all else you need to know that your backup strategy (including the specific data storage technology used, the software used to perform the backup, your media rotation strategy, and all the other factors that go into making your backup strategy work) will actually work to allow you to restore data. This means that you need to test your backup strategy under as close to real world conditions as you can to be sure that it really works. The time to test your backup strategy is NOT when you have a real need to restore data.

Backup should provide several basic attributes:

• Sufficient coverage to make a restoration efficient. While you might think that backing-up only your data files is necessary (because you can just re-install the application software) the trade-off in using less space on the backup media and less time in taking the backup is often not a "win" when you figure the time taken during a high-pressure restore scenario to re-install software that could have just been restored from backup. Assume that you should backup everything and think long and hard about why you would exclude something from backup.

• Sufficient frequency as to balance the cost of data loss with the cost and practicality of backup. A source that stores files for many users may have tens of thousands of dollars (or more) of worker-productivity invested in the files stored on it each day. More frequent backups will cost more, but the cost may be justified by the expense associated with data loss if the backups don't happen frequently enough. You need to understand, from the users who use the source data being backed-up, what the cost associated with data loss would be to the business and make the case for backup with sufficient frequency as to minimize that cost.

• Off-site storage to mitigate physical disaster. Keeping all your backups in physical proximity to the source data is a recipe for data loss during physical disaster. Floods, fires, thefts, etc, can all destroy the source data. If the backup medias are co-located with the source data then you risk destruction of the backup media at the same time as the source data. Be sure to think about confidentiality and integrity requirements in off-site storage strategy. Sending an unencrypted copy of your data home with an employee (who, invariably, leaves it in their car to be stolen) isn't necessarily a great off-site rotation strategy. Securing your off-site copies from tampering and unauthorized disclosure is just as important as having off-site copies.

• A history of multiple backups to allow future restores. It may be necessary to restore data from some time in the past, and keeping only a few (or worse, only one) generation of backup media is a terrible idea. Someone may not realize that they need data restored until days, weeks, or months after the data is lost. Having multiple generations of backup media allows for restoration in the future. This is another area where a trade-off of cost versus business need is very important. The backup should fulfill any legal or organizational needs the business with respect to retention of data for restore. (It's important not to confuse backup with archival here, either. They're orthogonal concerns that share some similar attributes and may use the same technology, but they're not the same thing.) In some cases, retaining too many backups can be a bad thing, so be sure that the business understands any legal ramifications for backup retention.

• Integrity of backed-up data. Part of testing your backups should involve verifying the integrity of the data that's restored. For some businesses this may be as simple as restoring a few files regularly and checking them out. For businesses with very strict security requirements this may amount to periodically sending a backup media to a third party for restoration and validation in a "clean room" environment. At the very least, you should be able to validate that mission critical applications and data are able to be restored with integrity preserved. An attacker who is corrupting your data may also corrupt your backup so that it reports that data has integrity when it doesn't. Be aware that keeping your multiple backup history in an "online" state may cause it to be subject to breaches of integrity by attackers. Some backup technologies (tape, for example) allow data to be kept in a totally offline capacity to allow physical barriers to help protect integrity (keeping the media under lock and key with a documented chain-of-custody).

• Backup and restore within an allotted time window. Your backup and, even more importantly your restore, should be able to be accomplished in an amount of time that satisfy's the needs of the business. An online backup, for example, sounds like a great deal until you find out, in an emergency, that restoring your 200GB of data will take 3 days of constant streaming of data over your DSL line. Keep restoration windows in mind when choosing a backup technology.

When you're evaluating specific technologies (disk-based backup, tape backup, online backup, etc) and specific backup strategies (grandfather / father / son media rotation, daily differential and weekly full backups, etc) think about how the interplay of software, hardware, and procedure work together to create solutions (or problems) that help (or hinder) fulfilling the requirements of backup. There are no blanket "right answers" and it varies for each business and, within a business, for different types of resources being backed-up.

Specific application software may have its own requirements that affect how to your run backups. Database servers (and database-like servers, like Microsoft Exchange) have their own peculiarities that you may have to contend with when designing your backups strategy.

Storage and filesystem features, like snapshots or live duplication, may help you perform backups of "live" data under tight windows.

---

Your specific needs

In the case of your specific application you describe you should probably look at how the built-in Windows Image Backup functionality can help you protect the operating system, configuration, and data. The built-in backup functionality isn't very sophisticated when it comes to policy and scheduling, though, so you may want to evaluate third-party software offerings (Symantec Backup Exec, CommVault, Bacula, etc) if you need a more sophisticated solution.

Talk to the people who store data on your server and find out what their tolerance is for data loss, what kind of retention windows they'd like to have, what their security concerns are, and what type of restoration timeline is required to help justify the budget for hardware, software, and labor. I'd strongly urge you to consider an off-site component, as well. Perhaps you'll find that using hard disks for off-site storage works well, but I'm personally of the opinion that there isn't an "enterprise ready" hard disk-based backup solution yet (one that's cost effective, reliable, and that operates consistently).

Answer 2

Or in short, for small businesses on Windows: use the built-in windows backup tool to write an external hard disk (preferably on an eSATA connection or in a SATA tray. The Windows backup tool can take advantage of shadowcopy to get copies of files even when 'they are open'. If this is of no concern to you, then you could even use robocopy to speed things up.

For disc encryption, you can use truecrypt: you can set it up so that it automagically opens up the encrypted volume when connected to a trusted machine (keys are the key :-) ).

Do read and think about the longer answer, because backups need to be thought about and managed, in order for them to be useful when they are needed.

Answer 3

Great answers so far. I recommend you look into Bacula and its guides and how-tos.