For more than 30 years, data tape in one form or another was the most logical choice for archiving media across various industries. Tape offered the capacity that early hard disks lacked, and although tape media and tape drives both had their fair share of problems, tape archives could be made secure by creating two independent copies of each file on different tapes.
The idea of replication did not translate to the hard disk world and RAID quickly became preferred for its ability to aggregate together the capacity and bandwidth of an array of small disks to provide a useful amount of fast storage. More importantly, it came with a self-healing capability to protect against disk failures.
Fast forward to 25 years and we have witnessed disk capacities increasing from 1 TB to 10 TB in a cat and mouse race against increasing data tape capacities. Today disks have the edge on capacity, but tape cartridges have the edge on cost.
However, the important question to ponder here is – for how long?
The demands of the cloud and big-data data storage are driving the cost and capacity of both. But it is reasonable to assume that a non-linear medium will ultimately be the winner.
Today’s high capacity 8 TB and 10 TB disks offer excellent cost and capacity. Nonetheless, two fundamental issues have to be addressed to create a secure and cost effective media archive.
POWER COSTS AND SECURITY
As recently as five years ago, popular opinion was that hard disks would never be able to compete with tape, due to the cost of the power they used and the cost of the air conditioning to get rid of the heat they dissipated. On the face of it, this seems to be an insurmountable problem, compared with a tape library, where the recording media is passive.
Turning to the security aspect, it became evident that RAID itself was the Achilles’ heel. In one form or another, RAID is still predominantly the technique used for online disk storage, albeit sharing the top tier with increasingly affordable solid state storage.
With any well-designed RAID, the risk of loss is very low when the individual disks are small, but the vulnerability increases with disk capacity. Even a dual fault-tolerant RAID can expose media assets to catastrophic loss for a window of up to 10 days. The moral of the story is – don’t assume that RAID necessarily means that it is ‘secure’. Clustered storage and Object storage get around the problem and as a result they dominate the Nearline market, neatly dividing the tier in terms of performance and cost, offering more choices for the system designer working to a budget.
Replication has also re-emerged for Online Cloud storage, and Offline Archives, providing the highest levels of security and unlocking the potential of even lower cost Shingled Magnetic Recording (SMR) disks. But here again the technology branches into two subtly different forms, which has a very big impact on Archive Storage.
Replication for the Cloud is based on segmenting the files and storing the segments in geographically separated locations. This diversity means the data will survive a disk failure or even an entire data center failure.
OFFLINE DISK ARCHIVES – AN ALTERNATIVE TO DATA TAPE
For broadcasters and media companies, the most important attributes of an archive are longevity, security, removable media and total lifetime cost of ownership (TCO). Non-segmented Replication ticks all those boxes by creating two identical copies of each complete media file, on different hard disks. These Offline Disk Replicas can be in the same enclosures at the TV station, or distributed between geographically separated locations on a wide area network.
Since the files are not segmented, individual disks can be removed for various reasons – be it shelf storage, transportation or disaster recovery. This technique also allows the disks to be fully spun down and stopped, with individual disks
ALTO is a simple storage implementation using non-linear disks to replace the linear tape media