Emerging trends in file-based infrastructures
Oct 1, 2007 12:00 PM, BY BRIAN CAMPANOTTI
The history of business is filled with examples of once dominant corporations that failed or were slow to react to technology change and lost their market advantage. Despite many indicators to the contrary, the leading players in these industries viewed themselves in a traditional way while the market around them was undergoing a fundamental shift toward a new business model.
Today, the trend in the broadcast world is toward divergent, fragmented audiences that demand relevant content on a multitude of nontraditional platforms. Content owners must efficiently and quickly evolve with the times to ensure future stability of their business. A mandatory evolutionary step in the broadcast world is to move away from traditional videotape-based workflows toward file-based infrastructures.
File-based implementations affect every aspect of the broadcast workflow from capture and ingest through post production, storage, distribution and transmission. Add to these considerations the delivery of multiformat and multitarget content, and it becomes a very complex puzzle to solve. It is essential that the platform assembled provides the greatest flexibility to adapt to emerging file formats as well as new modes of content distribution and consumption without placing additional burden on the human component of the workflow.
Figure 1. An advanced CSM-based infrastructure. Click on image to enlarge.
To accommodate the advanced requirements, state-of-the-art file-based infrastructures are built with CSM systems at their foundation. This middleware solution provides the integration and abstraction layer allowing high-level user applications, such as automation and media asset management, to communicate and transparently control the complexities of the vast underlying storage infrastructure.
CSM vendors now offer unified integration to encoding platforms, transmission devices, newsroom systems and NLE platforms. They also provide key functionality, such as transcoding and automated content lifecycle management. (See Figure 1.)
After A/V content is encoded into a file, a properly designed CSM-based infrastructure allows that content to be stored, replicated, repurposed, transferred and reformatted across high-speed local or WANs. It does this without any human intervention and at many times faster than real time.
A fundamental portion of the CSM solution is the file-based storage infrastructure. These high-speed, mass storage devices are constantly evolving, largely driven by the IT industry. For the broadcaster, a range of technologies are now available to balance performance, reliability, scalability and cost. Typically these storage technologies include fast SAN disk arrays for content-intensive applications; less-expensive NAS disk arrays, robotic data tape and optical storage libraries; and offline storage of media ejected from the robotic library but still actively tracked by the CSM system.
Disk arrays are fast and offer random access to content, but they do have significant downsides when parameters, such as cost per terabyte, scalability, cost of redundancy and disaster recovery, are all taken into account. In data tape-based robotic systems, linear access and mechanical characteristics (robot arm movement, tape mount times, positioning time, etc.) dominate. When comparing disk arrays with data tape-based robotic systems, it is obvious that intelligent analysis is needed to choose and balance the correct storage technologies when designing an effective CSM infrastructure.
LTO-4
The evolution of data tape technology continues to advance at an amazing rate and can be considered mandatory for any future-proof CSM implementation. The latest and most compelling is the release of the fourth generation of LTO technology, LTO4, supported as an open standard by multiple manufacturers. These tape cartridges have 800GB uncompressed capacity providing enough storage capacity for nearly 40 hours of IMX50 SD content. This represents a significant cost and real-estate savings over the same amount of content stored on traditional videotape stock. Even more impressive is the fact that each LTO-4 drive is capable of sustaining 120MB/s transfers, which equates to the movement of one hour of IMX50 SD content in a little over three minutes. This is considerable performance improvement over videotape-based workflows. Savings are increased manyfold when the human factors regarding videotape-based workflows, such as physical transport, replication and quality assurance checks, are also factored into the equation.
In a robotic library, 10 LTO-4 drives can sustain well over 1GB/s of sustained throughput and can scale from there depending on the content workflow demands as services and distribution evolve. These speeds are impressive, but are based on the prerequisite that the CSM solution itself can effectively scale to meet these staggering performance demands.
Robotic systems and optical media
The mechanical overhead and linear nature of data tape robotic systems cannot be ignored. However, they also do not factor significantly into the overall system performance of a properly designed CSM infrastructure. By designing a solution that includes a careful balance of nearline spinning disk augmenting a backend mass data tape robotic library, amazing performance can be achieved.
Optical media technologies, such as holographic and Blu-ray, continue to advance, but not at the same rate as data tape technology. Although they do exhibit random access performance closer to hard drives than to linear data tapes, their density, cost per terabyte and transfer rates still pale in comparison. These technologies better serve acquisition and front-end newsroom editing operations rather than long-term storage goals at this point, but this may change as technology advances.
Figure 2. Optical and data tape drive/library performance summary. Click on image to enlarge.
Mechanical performance characteristics for robotic, tape-based libraries can be methodically calculated to provide tangible performance metrics that can be used to factor into architectural choices. Figure 2 plots the end-to-end performance of different optical and data tape technologies in robotic library systems. Taken into account are robotic arm movement and media load times, media mount times, media seek times and drive transfer speeds for content encoded at IMX50 (or DV50) rates. The horizontal axis plots the duration of this file-based content in minutes (which directly maps to file size), while the vertical axis indicates the transfer performance as compared to real time.
For example, a plot point at the intersection of a 30-minute IMX50 asset on the horizontal axis and 6.00 on the vertical axis means the CSM system could transfer the content in a mere five minutes.
The main conclusion to draw from this multidimensional comparison is for short duration content (left side of the plot in Figure 2), the mechanical characteristics of the drive/library dominate the overall performance equation. As we move to larger duration items (toward the right side of the plot in Figure 2), drive throughput becomes the dominant factor.
| Want to use this article? Click here for options! |  |
