-
Framework:
- Start: 01/03/2007
- End: 28/02/2009
-
Project partners:
VRT
IBM
NextiraOne
Comsof
Porthus
UGent INTEC-IBCN
UGent ELIS-MMLab
UA PATS
VUB ETRO
The project has concluded.
Applying an ICT based infrastructure in video/media production introduces a number of possible fundamental benefits, facilitating the fundamental shift from traditional video manipulation to a file-based production paradigm. Technology enables video to be treated, processed, stored and transported as ordinary files independent of the video format, instead of the continuous streams used by the classical media technology of today. Amongst others, the most profound technology changes are:
- Server-based (non-linear) video editing or processing;
- Central disk-based media storage;
- Network-based access to the media;
- Software-based media management and media production systems.
Together with the appearance of some standards like MXF and AAF, which provide for a generic file-container for the media essence, these changes have led to the file-based paradigm for media essence. As a practical consequence, this has brought some of the major broadcasters to their ‘tape-less’ TV production vision. This idea is further supported by the appearance of camera devices with storage facilities other than the traditional videotapes, e.g. optical disks (Sony) or Solid State memory cards (Panasonic).
In the last few years VRT has, together with IBBT, conducted research on architectures for integrated file based media production infrastructure using standard IT technology systems. The fact that no media- or IT-supplier was capable of delivering an off-the-shelf solution for a central media storage infrastructure clearly demonstrated the value and necessity of such a research track and the importance of the leading role of a broadcaster such as VRT in this technological area.
The resulting architecture has recently been exploited in the production roll-out of a 400+ TB disk-based central FC-storage platform at VRT suitable for radio and Standard Definition (SD) TV production, and the ongoing adoption of the file-based paradigm for media production.
High Definition (HD) television is expected to become the standard video format within a few years. After taking the leap to file based video production, it is hard to imagine that broadcasters will take the step back to tape-based production in the coming High Definition environment.
However, these new HD formats introduce some formidable challenges concerning the media infrastructure, requiring the introduction of different technological components into the architectural model to cope with the increased bandwidth and capacity requirements in a cost-effective way.
Increasingly large file systems will be used for scalability and, above all, performance reasons. However, restoring the bulk of data from a secondary copy on data tape in the event of disaster recovery creates immense problems if a short restore window has to be maintained. With the advent of HD TV, quadrupling the storage capacity (and transfer bandwidth) requirements, the restore windows will grow accordingly.
The classical IT-backup/restore paradigms have to be revised and the tools studied and possibly re-tuned to see whether other technical alternatives are viable. Secondary disk based storage technologies are becoming less expensive and will have to be investigated as alternative to the prevailing data tapes used today.
As centralised resources gain in importance in an integrated end-to-end file based media production, more and more applications move from typical and specific media work centre applications to central generic media application farms. The ever increasing requirements for media processing and integration urge the adoption of the concept of GRID enabled computing as underlying technology for processor-intensive processes.
In a HD-based media infrastructure, emerging new media formats will supersede the current standard definition video formats (e.g., D10, DV, MPEG2), browse video formats (e.g., MPEG-1) and audio formats (e.g., MPEG-1 layer 2). The selection of appropriate media formats for the different aspects of a HD-based media platform is of great importance. In order to obtain an optimal balance between media quality, bandwidth requirements and efficiency throughout the different workflows, decisions on the selected media format have to be preceded by thorough consideration and assessments. Clearly, the upcoming HD media format war requires careful selection of the appropriate encoding format, taking into account the different production processes involved and their respective parameters.
The HD media production environment requires a complex, tedious and more flexible integration between the central infrastructure and its media asset management system, and the video and audio work centres since the open standards such as MXF and AAF are not yet fully matured nor adopted by every media solution vendor. Further refinement of the application of MXF and AAF in a media production environment is essential. This can be accomplished by defining extra application-specific restrictions.
On a higher level, this integration should be approached by introducing the concept of Service Oriented Architectures, or SOA. This defines the use of so called services to support the requirements of the business users. In a SOA environment, nodes on a network make resources available to other participants in the network as independent services that the participants access in a standardized way. This tends to avoid the chaos of implementing ad hoc point-to-point integrations between every couple of media applications.
This project aims at designing solutions that address the above bottlenecks by:
- Designing and building Proof of Concepts (POCs) of a High Definition media production and storage environment, based on an optimal choice of mostly IP based components. The project will cover research in the area of low end iSCSI storage, IP based SANs and disk connection networks, TCP/IP and iSCSI offload engines (TOE’s), IPoIB (IP over InfiniBand) and iSER as a possible alternative to TOE’s, different CPU technologies (AMD vs Intel), GPFS, GRID computing, cell-processors, alternative secondary storage concepts and new backup/restore paradigms;
- Studying new media formats, follow-up of standardization and evolution of MXF and AAF, Service Oriented Architecture (SOA), Enterprise Service Bus (ESB);
- Investigating the relevance of generic GRID computing concepts for the GPFS cluster based media infrastructure models.