Data recovery: Eliminate the weak links in back-up

Every company must have a comprehensive back-up and continuity plan to cover data loss, but checks must also be made on the back-ups themselves.

IT managers often wrestle with exponential data growth, data compliance requirements, and service and availability demands from the business.

As they implement faster back-up systems, they must make sure data is backed up properly and recoverable. In addition, the data generated by branch offices and mobile workers needs to be backed up and managed.

It is important to have a solid back-up and continuity plan, as well as policies for when and how back-ups are carried out, and which storage media are used.

Forrester Research analyst Lou Agosta said that enterprise back-up is crucial, especially if the company relies on a datawarehouse database.

“If the database is damaged for any reason and no back-up is available for executing a recovery operation, then lost revenues, angry – and lost – customers, and costly rework are certain. In the worst case, the company is out of business.”

Users can choose back-up systems from the most popular storage suppliers, such as Hewlett-PackardP, IBM, Hitachi Data Systems, StorageTek and Sun Microsystems, or opt for more specialist software or media suppliers.

In terms of the big enterprise back-up systems from the larger suppliers, the choice is huge, but there are many factors to take into consideration, such as service, media, scalability and data back-up and retrieval speeds.

Tape has been a popular storage medium for many years, for data archiving, back-up and disaster recovery. Tape can provide the bedrock of a company’s storage policy, as it is low-cost, ubiquitous and relatively fast.

However, low-cost disc-based storage is emerging as a complement to tape. “We are seeing a definite trend in storage and back-ups towards disc-to-disc technology. These devices go a long way to eliminate many of the problems seen with using tapes for back-ups,” said a spokesperson for analyst firm Frost & Sullivan.

For large businesses running storage area networks (Sans), the benefits of finding the right back-up and restore technology will mean the company will gain from reliable back-ups, smaller back-up windows and continuous business operations that do not have to stop as data gets backed up.

The systems available today can automate a lot of the back-up processes, saving valuable time, and can virtualise data storage, making efficient use of storage resources.

One example of a storage system for large enterprises is the Hitachi Data Systems (HDS) TagmaStore Universal Storage Platform. This consolidates and virtualises data into one pool, and can scale the storage up to 332Tbytes internal storage, and a massive 32 petabytes (1 petabyte = 1,024Tbytes) external storage – across multiple suppliers’ storage products.

The Universal Storage Platform also allows storage managers to create tiered storage configurations, which means data can be passed down automatically, over time, to less expensive and slower media – for example, from expensive discs to cheaper tape cartridges. This can be useful in meeting compliance regulations on data retention, and means that more current data can be accessed faster.

The Universal Storage Platform comes with the Hitachi HiCommand Device Manager application – a software platform for centrally managing, configuring and monitoring Hitachi storage systems as well as Sun Storedge and other products.

Sun has also dedicated a range of its large StorEdge storage family to back-up and restore, with media options including 36Gbyte DAT 72 tape cartridges, 160Gbyte or 300Gbyte SDLT tapes, or 200Gbyte or 400Gbyte LTO tapes.

Sun also provides tape libraries that can hold large numbers of tape cartridges. For most of these, the Sun Solaris operating system is a requirement, although some models also support Windows and major brands of Linux.

As well as tape, Sun offers StorEdge network-attached storage (Nas) disc-based systems, for fast back-up and recovery. The main product is the StorEdge 5000 Nas appliance running StorageTek File Replicator software.

Nas hard discs are set up with their own network addresses rather than being attached to the central computer that serves applications to network users. Nas, as a platform for consolidating storage, differs from a San, because it provides a dedicated system to back data up to, and has its own operating and server systems.

A San, on the other hand, can be based on Fibre Channel or IP (iSCSI). For Fibre Channel Sans, a separate network is required and consists of specialised hardware.

iSCSI Sans can use pre-existing networks, but are best designed as dedicated IP networks to avoid conflicts between storage and client/server traffic.

Iomega has also developed a Nas range, which uses its proprietary Rev removable cartridges. Storage capacity starts at 320Gbytes with the 200d Series, scaling up to 1.6Tbytes with the 400r Series.

There are options for hot-swappable Serial Advanced Technology Attachment (Sata) storage, to add capacity without taking the system down.

IBM’s Totalstorage line is a large range of back-up products and uses IBM Tivoli Storage Manager software to manage back-up and restore operations.

Totalstorage has many permutations, for example, to back up critical IBM DB2 database systems that support online transactions, analytical processing and mission-critical file servers. There is also a version specifically for SAP applications and databases.

An example of a branch office back-up system is Hitachi’s Truecopy Heterogeneous Remote Replication Software, which remotely replicates data from one Hitachi storage system to another. This helps keep the systems running and enables rapid recovery in the event of an outage.

“Moving data to a secondary site for switchover of primary processing can eliminate scheduled downtime and is critical when an unexpected event compromises your primary site.

Unlike traditional data recovery processes that are labour-intensive and can span several days, recovery based on Truecopy software can help businesses resume operations rapidly, almost from the point where they broke off,” said Hitachi.

EMC has also developed new products for distributed enterprise branch offices and small and medium businesses.

The EMC Clariion AX150 and AX150i networked storage systems use second generation iSCSI technology, and support the latest Sata II disc drives. These offer higher speeds and data availability than previous drives, and are more reliable.

Both products can be installed by the user, and scale from 750Gbytes to 6Tbytes capacity.

The AX150 allows firms to use Fibre Channel storage network connections, and the AX150i supports iSCSI. Both have their strengths, according to analysts, with iSCSI supporting Gigabit Ethernet, and the more traditional Fibre Channel having broad industry support.

There are many types of media, and options for backing up remote workers’ data. For example, it can now be done over the web using a hosted back-up service.

A more reliable way is to use a network hard drive – a small desktop hard drive with plug and share capability and wireless access. Iomega is one of the suppliers that offers such products.

Remote users are also able to store data on CDs or store their data on a number of removable media, such as USB keys, PDAs or portable MP3 storage devices such as the iRiver.

Benefits include portability and ease of access, with the drawbacks including security issues and the complexity of managing and centralising data.

In fact, as the number of mobile storage devices proliferates, the IT department will increasingly be required to control and secure the silos of information stored by remote workers.

John Monroe, research vice-president at Gartner, said, “As corporate data becomes increasingly mobile and accessible between now and 2007… dedicated storage teams will become a far more strategic element of the IT environment, data recovery will evolve as a central concern, archiving will assume a new role, and the use of data encryption as a means of preserving corporate integrity and value will become a standard best practice.”

How to implement an effective back-up schedule

Robert Ayoub, an analyst at Frost & Sullivan, recommended that companies make one full back-up per week, with appropriate incremental back-ups on all the days in between.

He said companies should ask questions such as, “Is it really necessary to back up all the system files of every machine, to determine the amount of time and number of tapes necessary to perform a full back-up?

“If a company is performing weekly full back-ups and daily incrementals and wants to make sure they could go as far back as a month reliably, then they should probably have a set of tapes for each week and rotate every month. Some companies want to store tapes off-site; some companies are comfortable with weekly back-ups.”

The benefits of only storing incremental changes rather than complete daily back-ups are faster back-up, and fewer tapes per back-up.

However, incremental back-ups can mean slower restores because you need all the tapes between the last full back-up and the current date, said Ayoub.

In addition, there is a higher risk, because all the tapes must be recovered since the last full back-up, and any errors could cause a failed restore. It is therefore essential to check back-up logs, he added.

If the back-up is infected by a virus or the data is corrupted then scenarios like these increase the importance of having back-ups going back to various points in time. If data corruption occurs on the only back-up set a company has, then there is not much they can do.

This is why checking back-up logs, retiring tapes before their shelf life expires and testing restoration of data from tapes is important, said Ayoub.

“Logs can be the first indication that back-ups are not completely finishing or that tapes are going bad.  Back-up jobs may look to have completed successfully but could have missed files entirely or could have failed checksums, and the administrator would not even know without checking the logs.”

Read more on Networking hardware