The preservation and protection of digital media assets has become a critical operational and legal directive for many businesses. For the broadcaster--whose dependence on disk systems for the capture, storage and play-out of their assets is growing--this trend is challenging on several fronts.
Planning and implementing protection schemes for these digital assets involves determining which methodologies, systems and applications are best suited for the workflow, and then balancing those decisions against risk and costs.
In planning for a media enterprise's digital asset protection, three principle categories should be addressed: the online or local protection for record and play-out; content backup, including critical data; and the restoration of that data should it be required.
For the general recording and play-out of disk-based content, modern server architectures are such that RAID drives and component redundancy seem to cover most protection issues for day-to-day online operations. Various secondary methods may be employed, including full mirroring of storage and server components, as well as offline, nearline and remote storage.
Those who began with video servers five or more years ago now find themselves having to upgrade. For those who did not initially employ fully mirrored storage arrays or other offline media replication, they may now be protecting content on a short-term basis by physically retaining the files directly on a supplier's cache or catch servers (i.e., the Pathfire, Vyvx or DG Systems cache).
This approach is risky because it places the control of that protection in the hands of a service provider. Often, facilities find that an unwilling deletion happened automatically, especially when the catch server is nearing capacity or the supplier thinks the material is no longer necessary to the end user.
DISCRETE INTERMEDIARY
The process of backing up data is complicated. One organization's idea of backup is often quite different from another's--creating incompatibility in third-party applications, such as automation. When it comes to setting a backup plan for critical data, users need to distinguish between a true backup and a mere copy.
Making a copy of the data still remains the only viable method for protecting it--but it's just one step in a successful overall plan.
The media content is sometimes viewed as the most important component to be backed up, overlooking the protection of the operating system's files, the metadata or the configuration files.
While many professional video servers already provide an internal redundant operating system drive for protecting the database layers of the system, their ability to replicate that data to another system is sometimes forgotten.
Automation systems may indeed provide a dual database for the day-of-air schedules and library lookup tables, but it is often up to the user to manage other third-party or non-automation data from a backup or protection perspective.
GROWING COMPLEXITY
A dozen or more years ago, backup practices consisted of simply copying all the data from a disk storage system onto tape. Incremental backups were done routinely, so data was considered protected. When the files were huge, recovery was often overlooked and seldom even checked for practicality.
Modern backup systems are now far more complicated. There are more choices in media, methodologies and applications. The solutions for total system protection involve more than just selecting a tape drive or a library robot and adding an appropriate backup application. In other words, backup and copying are not one in the same.
When backing up file systems, users must consider both the files and the associated metadata that accompanies those files. Metadata is not only that element that permits selective segments to be restored, but it is the key to the structure and meaning of the individual files and the volumes that are being protected.
Bulk-level backup is insufficient. Essentially a copy of all the data, a bulk transfer is when an entire bit-image of a disk volume is replicated to a secondary location.
Problems will surface in restoration, as the bulk restore is an all-or-nothing function. Bulk copies eliminate the ability to selectively restore segments; so as the file systems continue to grow in size, and complexity, so do the issues of backup and restore.
Historically, backup has been plagued with reliability, time constraints, control and even storage issues. Poor performance for almost all but the most exotic and costly of the tape library solutions remained a threat to reliability and value.
With today's proliferation of disk drive options such as ATA (Advanced Technology Attachment) and serial-ATA, the road is now paved for newer methodologies such as disk-to-disk (D2D) and even disk-to-disk-to-tape (D2D2T).
In data centers, D2D, D2D2T and DVD technologies are leveraging a multitier architecture that improves backup performance while mitigating accidental file deletion and partial or total storage failure. Yet even with these new disk application technologies, the strategies for backup and restoration must still include some fundamental understandings and balances.
MISTAKES IN MIRRORING
Mirroring was once thought as the ultimate protection scheme. However, one of the fundamental misunderstandings is that the mirroring of data on duplicate drive arrays is an alternative to backing up those files. Mirroring will not help protect data when a user (or another incident) accidentally deletes or damages a file, directory or volume, because deleting a file on a primary drive will generally delete the file on the secondary or mirror drive. Still, mirroring does prove practical if one of the drive arrays is double faulted, e.g., two or more drives in a striped RAID fail.
Another fundamental to identify is the restoration scheme, which should be routinely tested. As was learned during the Sept. 11 attacks, many companies had a backup plan with offsite disaster recovery, but the issue of restoration tested the true value of the plan--and in some cases the plans took an inordinate amount of time to invoke. Some cardinal rules for backup include testing every backup tape and developing a means to exercise the tape drives used for recording. The record and restore functions should be tested.
Routine and preventative maintenance is equally important. Organizations using datatape drives sometimes treat those data drive heads with only casual preventative maintenance.
Dirty or damaged tapes provide no value. Keep the tapes in their boxes, handle them with care and replace them if dropped or if known particles of dirt enter the container. The rules for videotape and datatape are the same. Keep the heads and the tape storage environments clean, cool and at a proper humidity level, and follow the recommendations in the manufacturer's specifications.
The longevity of any backup media depends on factors such as environment, age, number of passes, exposure to other magnetic fields and on rare occasions, poor formulation during manufacturing. Do not assume tape backup will be readable in four, five or six years. Tapes break down. Develop a practice to test and copy aging tapes to new media from time to time.
Critical tapes, such as master database files, operating systems and original software, should be stored on two identical copies. One of those tapes should be kept off site, and remember to inspect and replace those critical copies from time to time. A separate database of the versions, with the date of installation and any other pertinent notes or parameters should be maintained. Insist on reviewing backup procedures often and thoroughly.
GET IN SYNC
Finally, get the IT and broadcast technical staff in sync. Understand each other's needs and get cross-trained. Each entity will, at some time in their tenure, need to depend upon one and other. Develop a workable philosophy and revisit it often. Practice disaster recovery procedures and keep the systems well tuned.
Following these guidelines will minimize risks even when budgets are tight. Ignoring the issues will certainly result in disaster.
