Mathias Rosenthal - Fotolia
Backing up to the cloud is proving more efficient, easier to use and cheaper than tape backup. Using the public cloud also merges much of the functionality of backup with disaster recovery.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
To understand this merger, we need to look at the purposes of backup and DR. Backup is intended to create a copy of important files that is much less accessible to hackers than the working data, but is close enough to the data center that recovery of a file can be achieved quickly. DR is a means of storing a copy of that data far enough away that a natural disaster won't destroy both your data center and the DR copy. DR data is usually restored en masse, to enable a server image to be recreated, while restoring individual files is more common with backup versions.
The cloud has changed the boundary between the two functions of DR and backup, with implications for how data is moved, stored and managed. Correctly built, the same setup can handle both needs, relying on the distance to the cloud to provide the geographic dispersion necessary to obviate a natural disaster.
Getting started with cloud backup and recovery
The first step in backing up to the cloud is to select a software package. Most major backup packages include a DR capability, with a few adding the ability to create server, desktop and mobile clones very rapidly using ghosting, a process of taking stored images of operating systems, apps, data and settings that can be loaded onto fresh units. Ideally, for fast recovery, storing compressed, ready-to-load images is the best way to do DR.
Software can be installed on either a cloud instance or a separate dedicated server. Since most retrieves to a backup system -- as much as 80% -- occur on recent files, I lean toward the dedicated approach. That server can be provisioned with its own fast SSDs to speed the backup process and, also, by caching recent files, can provide a way to quickly recover many files without WAN or cloud delays.
Once you are comfortable with the software, which involves checking internet references and test driving the code, the next step in backing up to the cloud is choosing a vendor. This is as much an economic decision as anything, but there are a couple of caveats.
Pick a disk-based archive service from your cloud service provider (CSP). Tape archives have to queue up tape loads and data streaming, and this can add hours to each system restoration, even if only one device is being restored, making their use for backup, as opposed to archiving for DR, problematic.
Next, pick a service zone in your chosen cloud that is more than 1,000 miles away from your data center. I learned the hard way that natural disasters, such as hurricanes, can bring mayhem to huge areas.
Unloading and maintaining your data
The next step in backing up to the cloud is that first data dump. This can be done by sending everything over the WAN or by creating a set of disks or even tapes and shipping them to the CSP. In fact, the CSPs have this down to a fine art, with drive caddies and special trucks capable of carrying petabytes. The backup software tool will have compressed the data, usually in a 5:1 ratio or better. Good practice is, before compression, to separate the common parts of images, such as operating systems and apps, from personalized data or settings, which can save even more space.
The process of keeping information up to date while backing up to the cloud is more complicated. One alternative is like the traditional daily backup, but a tool that can update incrementally is much more sensible, since it limits data loss to a short window of time. Done right, this enables point-in-time recovery, which can help synchronize all the recovered servers.
Dealing with ransomware or any other malware usually involves a data restore from backup copies, but it is possible that these may be contaminated by some exploits. DR or archive copies are more remote as a result of compression and the use of software to restructure the saved data.
The data should also be harder to get to, and should not be available online to all users as a resource. If recovery from an attack fails, it might be necessary to clone new instances or systems and, there, the value of the DR approach really shines.
Five questions to ask potential cloud backup providers
Best practices for cloud backup and disaster recovery
Options for backing up data to the public cloud