Blog

Different Types Of Incremental Backup

Transcript:

With data that changes frequently, you’ll want to make daily copies of your backups and keep historical versions on file.  Although “backup everything” might seem like a good policy, keeping too much data can cause just as many problems as keeping too little data.  You could end up violating information compliance regulations, opening opportunities for data leaks, and creating unnecessary maintenance burdens.  So make sure that this isn’t just an IT project.  Bring other departmental decision makers and your legal team into the discussion and ask for their input.

Some of the things you’ll want to consider are how much data needs to be backed up, how often it needs to be backed up, how many copies of your data to keep, and what measures will you take in order to protect the data while in storage.  Once you have a clear idea of your company’s backup requirements, it will probably be determined that certain types of data require daily point and time backups that go back many years.  Of course, making a new full backup every day and keeping an unlimited number of copies on file would be too much.

A typical backup rotation involves making and retaining a new backup every day for a specified cycle and then archiving a snapshot for long-term storage once per cycle.  For example, you might create a new backup version every day for 40 days and then take a periodic archival snapshot of your backups on day 30, and the process continues from there.  However, this approach brings up two major challenges.  It eats up a lot of expensive storage media and it creates long daily backup windows where the servers can’t be used.

As a solution to these and other problems, incremental backups were created.  What do we mean by incremental?  Incremental backups are simply backups that exclusively focus on data which has changed since a certain point or time.  There are three generally accepted approaches when it comes to copying incremental changes for backup: file level incremental and block level incremental.

File level.  A file level incremental backup will automatically backup an entire file whenever it changes.  If you have a folder full of images and you edit one of them, that image will be the only one which gets copied.

Block level.  A computer file is stored on the hard drive as a series of data blocks.  The computer keeps an index of all stored files and the addresses of the blocks which make up those files.  It’s common for operating systems to break up a large file into thousands of blocks and them randomly spread those blocks out across the entire hard drive.  When a file changes, a block level backup will only back up the data block which has changed instead of copying the entire file.  If you were working on a 100-megabyte PowerPoint file and only changed a single slide, a block level incremental backup would be smart enough only to copy the data blocks associated with the slide that you’ve changed.

Today there are three leading methodologies when it comes to incremental backup.  They are cumulative incremental, differential incremental, and incremental forever.  To illustrate the differences we’ll use the example of a company with a 40-day backup cycle.

Cumulative incremental.  This approach involves making one full backup at the beginning of the cycle and then only backing up the changes that have occurred since that original backup.  If a full backup happens on Sunday, then Monday’s backups will only contain Monday’s changes, but Tuesday’s backup will contain the changes from Monday and Tuesday, and Wednesday’s backups will contain changes from Monday, Tuesday, and Wednesday.  Although the incremental backups on the second or third day might be fairly light, they will get significantly larger as you reach the 35th or 36th day and all of those changes add up.

At the end of 40 days, a new full backup is performed and the cycle starts over from day 1.  For companies looking to cut backup storage costs and minimize backup windows, this might not be the best approach.  However, the cumulative incremental approach does offer fast recovery since only two storage devices, the full backup and the most recent active version, are required for recovery.

Differential incremental.  This approach is similar to the cumulative incremental approach.  However, it only records incremental changes from the previous day.  Monday’s backup will only store the data that has changed since Sunday.  Tuesday’s backup will only store the data that has changed since Monday.  Wednesday’s backup will only copy the data that has changed since Tuesday.  At the end of 40 days, a new full backup is performed and the cycle starts over from day 1.  The differential incremental offers the fastest backup speeds and uses up the least amount of storage.  The downside with this approach is that recovery is much slower and more error-prone since you need all of the previous backup copies on hand in order to restore your data.

Incremental forever or synthetic backup.  Incremental forever, often called synthetic backup, is a software-driven approach that does away with the need for a full backup at the end of every cycle.  Instead, you just create a single full backup and copy your incremental changes over every day.  When this data arrives at the backup server, all of the changes are recombined and a new full backup is artificially created on the backup server.  This approach combines the speed and cost benefits of the differential backup process while also providing fast, simple recovery.

Online backup companies rely almost exclusively on the incremental forever backup process.  End users prefer this methodology because it’s completely automated, doing away with manual maintenance and costly tape storage and shipping.  Instead, companies can now set their servers to automatically encrypt their incremental changes and send them over to a remotely hosted data center.  In an emergency, this data can be quickly, easily, and securely restored.

This approach has also gained popularity in recent years since the elimination of periodic full backups helps companies more effectively deal with the problem of rapid data storage growth.  There are many reasons why your company would need to protect their data, such as disaster recovery or legal compliance.  Before you develop a backup plan, it’s important to understand why you’re backing up your data in the first place.  And over time, requirements will change and scope will grow.  For this reason, your data protection needs to remain simple in order to be flexible.

Thank you for taking the time to watch our video.  If your company would like help in deciding which incremental backup methodology is best suited for its data protection, please visit storagepipe.com for more information on Canada online backup.

Leave a Reply