When implementing data deduplication, there are some inherent challenges to take into account. The first is that although data deduplication reduces storage and WAN bandwidth, not all data deduplication is created equal. Each vendor has their own algorithmic approach and with the exact same data mix, backup size, and retention period, different solutions will achieve ratios of 2:1, 4:1, 6:1, 8:1, 10:1, 12:1, 20:1, and higher. Depending on the deduplication algorithm used, the amount of storage and bandwidth can vary greatly. A strong deduplication solution can achieve a deduplication ratio of 10:1 to as high as 50:1, with an average of 20:1 depending on the data mix and retention period. Typically, the backup application’s deduplication results in a lower ratio and uses more disk and bandwidth than dedicated target-side appliances, since appliances have dedicated high-speed compute and can use more aggressive algorithms to achieve higher deduplication ratios.

Another challenge is that most vendors simply added data deduplication as a feature to a scale-up storage platform or to a backup application. This approach of adding deduplication as a feature versus creating an architecture specifically for data deduplication not only slows down backups, restores, and VM boots, but as data grows, the backup window expands; data deduplication is highly compute-intensive and when this processing is performed in the data stream of the backups, backup speed suffers.

The impact of implementing an inferior solution may include:

• Lower deduplication ratios, resulting in 1) additional storage and cost, and 2) increased bandwidth and cost due to replicating more data
• Slow backups, resulting in long backup windows
• Slow restores, offsite tape copies, and VM boots, impacting users and productivity
• A backup window that expands as data grows, requiring the regular upgrade to a bigger and faster front-end controller or media servers (known as “forklift upgrades,” which are costly and disruptive)

When data is deduplicated during the backup window, the backups are slow resulting in a longer backup window. In addition, only deduplicated data is stored, so in order to perform restores and VM boots, data needs to go through a time-consuming reassembly or “rehydration” process each time. ExaGrid looked at these challenges differently than other vendors. They are not merely storage and replication challenges because data deduplication introduces a compute challenge that other approaches do not address.

ExaGrid’s hyper-converged secondary storage for backup approach meets all five challenges by delivering the:

• Best deduplication ratio for least amount of required storage
• Best deduplication ratio for least amount of bandwidth used
• Fastest backups for the shortest backup window
• Fastest restores, offsite tape copies, and VM boots to improve user uptime
• Backup window that remains fixed in length even as data grows

It stands to reason that as the volume of backup data grows, so too does the amount of data to be deduplicated, and it follows that if additional compute is not added, the backup window will continue to grow indefinitely. However, ExaGrid is the only solution that uses a hyper-converged scale-out storage architecture, so instead of adding just storage behind a fixed resource controller, ExaGrid adds full appliances with all required resources – processor, memory, network ports, and storage – to a scale-out system. Therefore, when data doubles, triples, quadruples, ExaGrid doubles, triples, quadruples all necessary resources, not just storage capacity.

ExaGrid is the only solution that meets all five backup storage challenges with the:

• highest storage efficiency
• lowest bandwidth usage
• fastest backups
• fastest restores, offsite tape copies, and VM boots
• fixed-length backup window despite data growth