Its no surprise that disk-to-disk (D2D) backup is being embraced by companies of all sizes. The most common method is using a storage system, either NAS or SAN, and setting up a policy for your backup application to back up to a disk target. A large number of companies have leveraged their existing storage systems using ATA/SATA drives to perform D2D backup. Additionally, appliances that are specifically designed for D2D backup are...
available from various vendors (e.g., Data Domain, Exagrid and NetApp).
D2D backup software products are also available that actually replace or augment your existing backup applications. Asigra and Avamar offer backup software applications that use disk storage systems as their backup targets. Both of these software products are easy to use, and provide high performance in addition to very compelling value.
While D2D is becoming a requisite part of the backup process within local data centers, the next phase will be to move to D2D2D. The concept of D2D2D includes using some form of remote replication capability to actually make copies of data stored on D2D backup devices to an offsite location for DR purposes.
This is valuable for a number of reasons. Compared to using a primary storage system, a D2D backup product can be an inexpensive and easier-to-manage alternative for DR when performing remote replication. Performing remote replication using D2D backup enables companies to make DR copies of ALL of their data via a single platform. Performing remote replication from a single storage system only protects data on that particular device. However, companies can back up data from multiple servers with internal storage, as well as NAS and SAN systems, to a single D2D backup product.
Since all of a company's data can be backed up to a single D2D backup product, it can then be replicated to a remote site for DR. This is a much simpler and more cost-effective approach than implementing remote replication on multiple SAN, NAS and DAS systems. A D2D backup product can provide a centralized way to protect primary data stored on heterogeneous storage systems.
The one trade-off of this approach is that recovery point objectives (RPO) are typically going to be 24 hours since backups will usually be done once a night. Compare this with synchronous mirroring, with a theoretical RPO of zero, or asynchronous mirroring, with an RPO that ranges from seconds to hours. However, a large number of companies are still doing nothing more than backing up to tapes, which are then vaulted offsite for DR. Utilizing a D2D backup system with remote replication that provides an RPO of 24 hours offers greater protection than most companies have today with tape backups and offsite tape vaulting.
An important enabler for D2D2D is data de-duplication, a technology that stores only unique backup data and can result in a 20-to-1 reduction in capacity. The impact of data de-duplication is landscape changing. Companies can store 20 terabyte (TB) of backup data on just 1 TB of disk capacity. Enterprise Strategy Group (ESG) Lab has tested products that perform data de-duplication and can confirm that it really does achieve 20-to-1 reductions. However, this type of reduction is typically achieved as part of an ongoing process. Additionally, while the process of removing duplicates can significantly reduce capacity, even greater results are realized when combining data de-duplication with compression.
Using a product that supports data deduplication will result in an initial full backup with approximately a 3-to-1 or 4-to-1 reduction. Incremental backups may double the ratios to a 6-to-1 reduction or better. While these ratios are impressive and will reduce costs, the next full backup will then achieve the more compelling reduction ratios. In ESG Lab testing, we experienced up to 60-to-1 reduction at this stage, but if we had continued on to perform weekly backups, over time that reduction would probably even out to 20-to-1.
Another technology that offers efficient movement of data over remote sites is something called differencing. This approach sends only changes or delta data over the WAN. Differencing can also provide amazing capacity savings, as well as enable WAN efficiency in the same ballpark as data de-duplication.
Companies can back up data from internal servers as well as from DAS, NAS and SAN systems, across heterogeneous server and storage platforms to D2D backup systems and then replicate them over long distances (D2D2D). This provides an easy, cost-effective and more fully encompassing DR product than standalone storage system-based remote mirroring. Companies still using a tape-only strategy should look at D2D2D. Additionally, D2D2D can complement storage system-based remote mirroring products -- allowing you to protect your mission-critical data using storage system-based products and D2D2D for the rest of your data.
Another major benefit of D2D2D is that it can significantly reduce and potentially eliminate the need to vault tapes offsite. Don't reject this notion out of hand. While most large companies may not have the option to do this, small and midsized companies are much more open to this idea. At the very least, it will allow these companies to minimize the number of tapes stored offsite. We are seeing that today in a big way. Many companies, large and small, are changing their backup processes by performing daily incremental backups to disk and monthly backups to tape. Since these companies have been using D2D2D, they also keep remote copies of backup data on disk, in case major local failures occur. Expect D2D2D to become a rapidly adopted strategy going forward.
About the author: Tony Asaro is the senior analyst for Enterprise Strategy Group.