This article can also be found in the Premium Editorial Download "Storage magazine: FC, iSCSI, NAS: How to choose storage for virtual servers."
Download it now to read this article plus other related content.
Hybrid vs. pure cloud backups. In a pure cloud backup scenario, agents on protected servers and desktops perform backups directly to the cloud. Quick setup and minimal maintenance are benefits of this service. A pure cloud backup product is best-suited for personal backups and backups for smaller firms with limited amounts of data to protect (typically a few terabytes). The drawbacks of backing up directly into the cloud are performance and bandwidth challenges because of latency and bandwidth limits of available Internet connections; these shortcomings are most important when restoring data.
Latency and limited bandwidth are mitigated by hybrid cloud backup products that use an on-premises disk or gateway as the initial backup target from which the data is replicated to the cloud. The on-premises intermediary usually caches the most recent backups for on-premises restores, minimizing tedious recoveries from the cloud; it also moves data into the cloud asynchronously. For a pure cloud backup solution without the on-premises intermediary for quick restores, it’s essential to understand all restore options, including the ability to have backups shipped to you on a disk or NAS device; restore options become more relevant as the amount of data stored in the cloud grows. Similarly, some MSPs accept the initial full backup on an external storage device (known as “seeding”) to avoid a time-consuming first backup over the Internet.
Compression and source-side deduplication are two technologies that help minimize the amount of traffic sent into the cloud. Data deduplication reduces bandwidth usage and also helps cut the cost of backing up to the cloud. Because cloud storage pricing is usually based on gigabytes stored, compression and dedupe are instrumental in lowering monthly fees. To maximize data reduction, some MSPs deduplicate on the source side and one more time in the cloud. While the scope of source-side dedupe may be limited to a single or few hosts, dedupe in the cloud can be performed against all data, resulting in significant additional data reduction.
CLOUD BACKUP CHECKLIST: KEY FEATURES
Enlarge CLOUD BACKUP CHECKLIST: KEY FEATURES diagram.
“We deduplicate and compress before we send data across, and we deduplicate one more time once data is in the cloud,” said Karen Jaworski, senior director of product marketing at i365, a Seagate company and backup MSP.
Transport. Besides source-side dedupe, cloud backup products differ in the way they manage available bandwidth. The ability to limit and throttle bandwidth while backups are in progress helps minimize the impact on users and other apps sharing the Internet connection. Moreover, being able to configure multiple bandwidth limits for different times of the day helps optimize the balance between backup performance and the impact on other users. Some cloud service providers, such as AT&T, give customers the option to use a multiprotocol label switching (MPLS) circuit instead of the Internet; this option is relatively cost-effective for customers who already use MPLS. The quality of service (QoS) feature of MPLS lets users label backup data as low-priority traffic, eliminating the impact on other users and applications altogether. This is especially attractive for midsized and large companies with many users and a lot of protected data.
This was first published in September 2011