Back in 2004, the drop in SATA HD prices didn’t go unnoticed. At the time, I managed a small data center and quickly became fed up with managing tapes. Backup windows were too high and physical tape management was tedious.
I decided to purchase an inexpensive Promise direct-attached storage (DAS) array and backup directly to disk. The Promise array held approximately 15 disks, providing about 4.5 TB of raw disk space. I made the array a target for my ARCserve backup software. It increased my backup times and allowed me to archive to tape during the day.
By today’s standard, that solution was a hack. However, it was a preamble to tapeless backup. Now, it’s worth taking the time to examine the modern day options for tapeless.
EMC Data Domain
One of the biggest challenges with HD-based backup is the cost per GB backed up. It is extremely difficult to beat the price of tape on raw cost per GB. While backing up to disk increases speed for both backup and recovery, HD backup is expensive. Back in 2009, EMC purchased disk backup company Data Domain. The Data Domain product has dominated the enterprise disk-to-disk backup market, with its primary advantages being deduplication and replication.
Deduplication reduces the total amount of physical space required for repetitive datasets. Backup is a perfect use case for deduplication. There’s ample repetitive data even when using incremental backup schemes. Deduplication has the potential of expanding usable disk space by as much as 30 times. The second advancement is replication. Data Domain allows for shipping backups off site through efficient replication.
EMC also enables direct backup to nodes via NFS and application integration. For organizations needing full-featured schedulers and backup agents, Data Domain serves as a backup target for many modern backup applications.
Two curiously similar approaches from Rubrik and Cohesity offer a modern take on disk-to-disk backup. Both solutions were born out their company founders’ relationships with hyperconverged infrastructure leader Nutanix. The general concept is similar to hyperconverged infrastructure, in that x86 compute is used to create scale out storage infrastructures. The products offer many of the features in the Data Domain product, but in a scale-out design. Cohesity’s focus is on secondary storage while Rubrik is focused on data protection.
A key feature is the concept of clustering. New scale-out solutions leverage shared filesystems, enabling clustering. By implementing a clustered file system between nodes, administrators can add or remove capacity as needed.
Key differences between the two solutions are the approaches to integration. Rubrik is a complete replacement of your backup and, potentially, your DR solutions as well. One of Rubrik’s strongest selling points is the interface and workflow. Rubrik is designed to configure backup and replication directly from the application. Rubrik also supports running VMs directly from the appliance to reduce recovery time objectives.
Cohesity can work as a standalone data protection system. However, it’s better suited for integration with either Veeam or Commvault. This option may appeal to customers wanting to extend the capability of an existing platform used to backup legacy infrastructure. In the case of Commvault, customers can use the same interface to schedule and report on jobs that backup Unix or a virtual environment.
With more targeted solutions entering the market, it’s safe to say that the tapeless environment imagined years ago has arrived. Some organizations may find non-technical challenges around the logistics of tapes, and most modern IT environments have the capability to support a tapeless data protection strategy.