Using Blu-ray for corporate data storage is a relatively new idea and won’t necessarily be a good storage method for all businesses. So why would Facebook choose this technology over more established ones?
It has been reported that Facebook is testing an optical Blu-ray disk library in the first instance to store its compliance data. Blu-ray supports both re-writeable as well as WORM (write once, read many times) media. Data on WORM discs cannot be modified – only the disk itself can be destroyed – and WORM media types are therefore ideally suited to storing information that must be maintained in its original state, such as compliance and regulatory data.
The downside is that capacity once used cannot be reused and so it is unsuitable for any data that is required by regulation to be able to be deleted (user accounts and content) and is by definition always used to store inactive data.
Facebook is initially looking at using 1PB of capacity for cold storage – data that can be filed away that doesn’t require accessing on a regular basis – which for the social network includes duplicates of user videos and photos kept for backup purposes. In terms of suitability for the task, Blu-ray potentially passes the test. But what about costs?
The Blu-ray prototype contains approximately 10,000 optical discs and a petabyte of data in a rack-sized cabinet. In the following table, you can take a look at the specifications and approximate costs of a Blu-ray system versus an LTO tape library solution. Through this analysis, we found that tape can actually cost up to 95 per cent less than Blu-ray and at the same time take up half the rack space needed for the same storage capacity.
N.B. This table of analysis is based on the following assumptions:
- The cost of the sheet metal, power, fans and robotics required for one rack is assumed to be about equal between tape and Blu-ray
- The cost of the media is based on Internet research to find approximate lowest online pricing
- The cost of the drives is not included because pricing is so variable depending on the source and in the big picture, the drive cost is dwarfed by the cost of the media
As the table also shows, we found that tape transfers data up to six times faster per drive and 20 times cheaper than Blu-ray. While our analysis is based on some assumptions noted above, it is very clear that tape has a clear cost and performance advantage for storing data for long periods of time. Tape may be one of the oldest forms of technology, but tape innovation and large volumes of data growth in the market have resulted in an increase in the demand for tape in large-scale, long-term archives.
The benefits of tape storage
Advancements in media and tape library technology have improved tape’s ability to maintain data integrity, together with increases in the archival lifetime for secure storage in an easily maintained environment.
- The development of LTFS (Linear Tape File System) has allowed tape to become self-describing, carrying its own file system metadata and thus becoming an open, portable format that can be read by multiple applications or using simple, free software
- Tape is extremely power-efficient and the enhanced power management capabilities of modern tape library systems provide greater operational efficiency, diminishing tape power usage when in an “idle” state
- Digital Speed Matching (DSM) functionality of tape drives match drive speeds to incoming host data rates; essential for the high performance drives of today, this technology has abolished the reliability issues caused by “shoeshining” in the past
- Tape is secure – inbuilt, inline encryption has no performance penalty and tape libraries can contain built-in key encryption management for no additional cost and great simplicity
- Tape can be integrated into a cloud storage infrastructure, addressing the challenges of bulk data movement and long-term storage costs
- The non-proprietary format of LTO tape solutions enables ease of integration and long-term deployment, contributing to tape’s case as a scalable, long-term solution
- WORM tape media is also available, enabling tape to be used for compliance applications and infrastructure to be shared with other data protection and archiving applications
Even if you assume Facebook builds its own racks and negotiates pricing for 100GB media that is closer to the cost of 50GB media – the point remains the same – tape delivers significant advantages in cost, performance and floor space. As far as power consumption goes, tape and Blu-ray are about the same because data is stored on media that doesn’t require constant power.
Furthermore, Blu-ray is a consumer-grade technology and is highly likely to cost more to service and support. For example, the Blu-ray proto-type is much more likely to incur errors, jammed drives and failed drives than an enterprise class, proven tape storage system.
We can draw from this information that Blu-ray is going to be a comparatively expensive option for Facebook moving forward. However, there are big name vendors such as Sony and Panasonic that have recently announced their intention to offer 1TB optical Blu-ray disks, so it will be interesting to see how Blu-ray storage evolves.
For now, it remains a medium mostly used for consumer storage, designed without the reliability associated with traditional big data archiving systems and at a price that will prevent major uptake in big data environments in the near future.
Conversely tape, with its progressive capacity growth, faster performance and lower power needs, has proved itself to be a serious contender when it comes to staying ahead in the data growth game.
Steve Mackey is the vice president of international at Spectra Logic