Data backup and disaster recovery made simple

The disaster recovery silo

Disaster recovery (DR) has several components. The first is moving data from the primary data center to the secondary data center. Traditionally, this meant backing up data to tape and shipping tapes off-site. Or, it might mean using replication software on the host, the array or the disk backup appliance to intelligently copy data to a secondary site over a WAN.

Many data centers used replication, but it had to be man­aged separately because the backup software was not inte­grated with it. So, the software had no way to verify that data was securely off-site.

Data backup and disaster recovery integration is hap¬pening in several ways.

The administrator had to manually and independently confirm that data was both protected locally and safely stored off-site.

Today, DR is being integrated directly into the backup process. Instead of being a separate product requiring sepa­rate management, disaster recovery is becoming a feature of the backup product. Data backup and disaster recovery integration is hap­pening in several ways.

Backup software as a DR manager

Some modern data backup software products can manage the data rep­lication processes mentioned above. For example, these products can initiate a storage array-based snapshot, back up the snapshot locally and instruct the storage system's replication software to replicate that snapshot to the remote location.

Other backup software products can back data up to a disk backup appliance and then trigger the replication pro­cess from the appliance. In both cases, the backup software, because it is in control, is fully aware of the quality of data protection both locally and remotely.

The downside is that there is typically narrow hardware support for these backup software products. For example, a large vendor that has hardware and software in a portfolio will only support the production storage or disk backup appliance that is within that portfolio. This single vendor support means that the organization needs to stan­dardize on that vendor's products.

Backup software as the DR tool

Alternatively, some backup software can replicate data natively. These applications use changed block tracking or source-side deduplication to provide a WAN-efficient method for backing up the application remotely.

Backup software that offers replication provides more flexibility than traditional backup software because it al­lows IT planners to leverage a mixture of primary storage and secondary storage devices. The downside of this ap­proach is that the organization must standardize on a single product to eliminate the other replication silos. There also may be some loss in performance, since the backup server has to drive both tasks instead of offloading replication to storage hardware.

Cloud backup is another option to ease data backup and disaster recovery integration.

Data movement is the first step. But, a complete DR strategy also needs to include restarting applications. Vir­tualization helps a great deal by lowering hardware costs in the DR site, but the IT planner has to balance cost savings versus performance during the actual disaster. Perfor­mance in the recovered state is critical. If there is a disaster, the IT planner has to make sure there is enough compute and storage performance to deliver near production perfor­mance for an extended period. This can require a significant hardware investment.

Integrating DR with cloud backup

Cloud backup is another option to ease data backup and disaster recovery integration. Using the cloud as a backup target has become very popular over the last couple of years. It can come in two forms, either a traditional backup product that offers the ability to use the cloud as a backup target or a cloud-integrated offering. Legacy backup plat­forms typically leverage deduplication and compression capabilities to copy data to a secondary site.

The same applies when using the cloud as a target. Some products use a gateway that converts a traditional NAS pro­tocol like CIFS or NFS to a cloud protocol like Simple Stor­age Service (S3) or CDMI. Others have incorporated native Amazon S3 capabilities into their software. The primary advantage of the native S3 integration is that it eliminates the need for a separate appliance.

Native cloud backup products come in two forms. The most common is a local backup appliance that can replicate data to the cloud using software. This allows users to retain a local backup for fast restores while sending older data to the cloud.

Alternatively, there are cloud backup applications that protect directly to the cloud via an agent installed on each server. Direct-to-cloud backup allows for dozens of systems to send data to the cloud simultaneously but requires in­stalling an agent on each system. It also does not retain an on-site copy for faster restores.

In either case, a cloud backup does fulfill the basic first step of DR: Getting data to a secondary site. Cloud backup is most appealing to organizations that don't have an "IT quality" secondary location. For those organizations, the cloud provider is that location.

A final decision to consider is what type of cloud will store the organization's backup data. There are two ba­sic options: a general-purpose public cloud provider or a purpose-built cloud for backups. A general-purpose public cloud has the advantage of scale, but it may be difficult to get specific support since the backup software provider is not in direct control of the cloud hardware. Purpose-built cloud providers have the ability to fine-tune the environ­ment specifically for their software and use case.

Storing a copy of backup data in the cloud fulfills the first order of DR: getting data out of the data center. It does not, however, fulfill the second: rapidly recovering a destroyed data center.

When recovering all data back to the primary data center, after a disaster, technologies like deduplication provide no value because a base set of data for comparison is not available. IT professionals considering cloud backup should also consider how they will move large amounts of data back to a DR site. Some cloud providers will ship data back to customers on an appliance -- typically at an additional cost.