Using Storage Area Networks to improve the restore times from backups

Steve Mackey, director of product marketing, ADIC.

With natural disasters in every newspaper, IT directors have a daily reminder that they must have robust backup and recovery measures in place.

Many businesses are so dependent on IT that if they lost their data, the business would be wiped out. IT directors are asked to think the unthinkable and create a solution that will get the business back on its feet quickly.

The targets can be tough: The recovery time objective (RTO) - setting how quickly the business must be operational again - can be as short as minutes for busy online traders for whom every second wasted is a sale lost.

Companies also need to decide on a recovery point objective, setting the maximum age of the data when it's recovered, and so the maximum amount of data lost.

It's not a decision they can take lightly: data loss costs time and money to correct but high performance data protection can be dear.

The best performance is achieved through synchronous replication, where data is updated in two arrays as transactions are carried out. Most companies cannot cost-justify this and many find that 24 hour tape-based backup cycles are the best fit for their data protection and budget requirements.

The problem with using tapes for disaster recovery measures is that although they're tucked away somewhere safe, they're not immediately accessible when you do need to restore from them. While the business braces itself for disaster, it rarely comes.

User error on the other hand, including accidental deletion, is an everyday occurrence and is responsible for nearly all restore requests.

If the user needs a file from the tape after it has been sent for secure storage, it can take hours for the tape to be recalled and for the file to be restored. Productivity can dive while users are waiting for data to be returned to them.

The textbook solution is to create two copies of the data - one which can be taken off-site for disaster recovery purposes, and one which can stay on site for user-driven restores. But with ballooning data volumes and extended business hours already causing tension, there isn't room within the backup window to create duplicate or clone tapes.

Now companies are improving their restore time by introducing disk to their backup process and using a storage area network (SAN) to make that possible.

In recent years there has been a trend towards consolidating storage using SANs. Instead of having storage devices attached to each server, storage devices and servers are connected to a storage area network.

The SAN's high performance fabric enables multiple computers to share the same disk array and enables administrators to allocate additional capacity only when required.

It also enables storage to be scaled more easily, eliminating much of the expense associated with over-provisioning storage, a common defence against space running out when using direct attached storage.

Consolidating storage provides management benefits too: analyst firm IDC estimates that storage administrators can manage up to nine times more data in a consolidated SAN environment than in a direct-attached configuration. Restore times are improved because there are fewer systems to manage and replicate from.

The SAN's greatest contribution to improving restore times, though, has been to make it possible for there to be a cost-effective way to introduce disk into the backup process. Even low cost SATA disk storage can cost six times more than tape storage, making it too expensive to attach to every server.

Attaching disk to each server also wastes capacity because it ties unused space to a server that doesn't need it, while other servers might be running short.

If plain disk is introduced into a multiple backup server environment, even in a SAN, this is exactly what happens, because a SAN on its own does not provide for sharing or 'virtualisation' of disk resources. 

If disk storage is attached to a SAN as a shared resource, for example as a virtual tape library (VTL), every server is able to leverage the benefits of disk while disk space is optimised for the network as a whole. Because space and read/write bandwidth is dynamically allocated as it is needed, there is no resource wasted through being dedicated to servers that don't need it.

Companies can use VTL to quickly perform an initial copy of data over the SAN and can then copy data from disk to tape outside the backup window.

This enables a marriage of the high speed and availability of disk backup, and the portability and disaster recovery properties of tape.

More importantly, it solves the puzzle of enabling files to be rapidly restored for users (from the disk) while also ensuring data is securely protected for disaster recovery (on tape).

By way of example, the ADIC Pathlight VX is a virtual tape library system with an integrated path to the tape library. This means that the secondary copy process to tape is done 'off-line' where there is no impact on the SAN or application servers.

With a total bandwidth of up to 2 TB/hour available, the initial backup process is completed rapidly and Pathlight's efficient use of tape resources means that far fewer physical drives are required.

Most importantly, a copy of data can be left on disk, so come the time to recover a lost file, there is no delay hunting down tapes before the restore can begin.

Disaster recovery imperatives have long decided how data should be stored, resulting in inefficiencies in day-to-day work.

The benefits of introducing disk are apparent every day, when somebody somewhere in the organization needs to recover a file that was deleted through error or a local hardware failure.

It's often hard to measure the increase in productivity resulting from being able to deliver such data quickly on demand, but most companies find that their investment in disk starts paying for itself immediately.

This investment is only possible because the storage area network consolidates storage and data, making the SAN a good first step for organizations that want to optimize their file restore times.

Steve Mackey is ADIC's product marketing director for EMEA with responsibility for introducing new products and technologies. He has over twenty years experience in technology marketing, ten of them in storage and seven of them with ADIC.