Weigh these factors when building a backup plan for your data center.
Data backup is an essential part of data center operations, but it’s important to really understand what makes a backup strategy successful. Most people say that it’s necessary to have a second copy of data in case the original copy fails. That’s true, but it is only part of the story.
A good backup strategy is obviously going to create that second copy, but it is more crucial that, when file recovery is needed, the data can actually be found quickly. This was a problem with straight tape backup, which could have recovery times in the tens of minutes for a single file.
Retrieval problems can be avoided with the correct backup software and the proper storage medium. The two are somewhat intertwined. Backup needs disk/SSD storage in the data path, allowing indexing of files to speed retrieval. That storage also buffers transfers to slow media, such as the cloud or magnetic tape. In addition, disk storage allows compression and deduplication of data before sending it to the final backup medium, saving transmission times and reducing the cost of backed-up data by roughly a factor of five.
The storage in the backup path can serve as a cache (with the right software, of course). With most retrieval requests occurring within 30 days of backup, caching offers a substantial improvement in quality of service.
Good candidates for an enterprise backup solution are disk-to-tape and disk-to-cloud packages. Tape and cloud come with different security risks. Tape is essentially an offline process. Alerting data requires manual steps and offline access. That’s a good solution for protection against hackers and, more importantly, the internal rogue admin or programmer. Tapes can be stolen, but good inventory control by your storage site should warn you of that and, with encrypted data, you are just out the cost of some tape media.
But that offline status is a real impediment to recovering individual files that have been corrupted or accidentally erased. Getting to that data may take hours or days and then it takes a long time to find the file. Backing up data to the cloud is very safe if data is encrypted at the source, and recovery time can be in seconds. However, the fact that it is actively online means deliberate or accidental data corruption is possible.
Now let’s look at what we store. More than 50% of data in large companies resides on mobile gear. A solid backup strategy must handle that reality. That means backup software must include policy-controlled backup of mobile devices.
Moreover, many companies are moving to the cloud for mainstream computing. The best solution for this data is to save it in a cloud backup or archiving service. This reduces cost and time for backups.
What to backup
Most enterprise backup strategies either just protect critical data or take a save-everything approach. It is better to consider recovery. What is saved should be enough to start the business from scratch. Obviously, all legal requirements must be met, including encryption key ownership, but the result will fall somewhere between the two extremes of critical data or all-in backup.
For example, keep copies of VM images, but if they're repetitive (and many are) deduplicate them first. This simplifies recovery and prevents generation issues. Recovery is best achieved if data sets are segregated wherever possible. That allows whole servers to be replaced quickly.
(Image: Matej Moderc/iStockphoto)
The security question
There is a marked security difference between (offline) tape backup and cloud backup solutions. Backup to tape achieves that desired disconnect between active systems and backed-up data. The problem is very slow recovery. The ideal solution is to use a different medium for remote storage. Google aims to do that with its Nearline service, but the very fact that access is fast means that security, in the sense of a completely separate offline storage solution, has been significantly reduced.
The issue is one of a physical firewall versus a reasonable recovery time. In the end, the recovery time should win, but it’s a dangerous victory if the user doesn’t take serious steps to isolate backup access from the main system admin, programmer and user pool.
Where to back up?
Public cloud backup creates both standard backup and disaster recovery, a clear benefit. Private cloud backup or local tape systems offer much less for disaster recovery. In many ways, moving to the public cloud for backup seems inevitable.
Disk-based systems, such as we see in MAID (massive array of idle disks) solutions, for example, are much faster in recovery, but could miss on the valuable firewall that making data offline brings. This has to be remediated by access control and business processes.
Much of today’s data is on mobile devices. You need software that handles that well. There will be a significant part of the remaining data in the public cloud, so the best place to back that up is in the cloud, too. Pick a versatile software package that recognizes these two realities.
Ease of use is important, especially on the recovery side. Having a 30-day local cache feature is a big plus, since recovery speed for most files is near instantaneous.
Encryption at the source server, with multiple keys owned by the backup team, is essential. Failure to do this could lead to undetected, successful attacks.
Compression and deduplication help reduce transmission time, storage space, and time to reload dramatically for most data sets. These features would be implemented on a local storage pool that is either on the server where the backup gateway is installed or on a networked space.
A tool supporting multiple parallel backup operations is useful in reducing backup windows, while WAN optimization can ensure data transmission at low congestion times or at lower connection tariffs.
Good recovery tools are crucial to a good backup solution. Check out the ease of recovering files, folders and volumes. Make sure the GUI has the ability to control user access and privileges.
Roll-back snapshots help recover data to a known stable level and allow enterprises to implement a policy of never deleting data. This policy may be cost-effective when using compression and deduplication, especially with mobile workspaces, virtual desktops and heavily replicated use cases such as web-server farms.
The arrival of a bunch of cloud-based newbies in the backup and archiving space is changing the price profile for backup solutions. Traditional backup is expensive at large scale, while the new players offer attractive pricing. With some of the cloud backup players now well established with feature rich software and large customer bases, traditional vendors are somewhat under siege as the realization that the game has changed sinks in with the customer base. Buying best-of-breed is no longer synonymous with paying high license fees!