Original article in TVB Europe (July/August 2019)
Written by Craig Bury, CTO, Three Media
Having to tell your management or your board that your media workflows or processes are not fully optimised and might be haemorrhaging money, and you need new tools to address this may not be the conversation you want to have.
And this is the challenge Three Media has faced over the years working with clients who know they need to introduce change to ensure success and competitiveness. We achieve this by introducing long standing principles of ‘describe once, exploit many’ and ‘management by exception,’ supported by configuration driven automated workflows and processes. These core principles have developed over time as have the products based on these.
Three Media started out 20 years ago as Broadcast Projects International, whose first large project was to help build the Digital Media Centre (DMC) in Amsterdam, which following its recent acquisition by TVT Media, became Europe’s first fully virtualised ‘private broadcast cloud’ originating channels for major network owned brands, globally.
The DMC project required several years of focussed effort, and by that, I mean that BPI was instrumental in the system and workflow selection and design, technology integration and implementation, testing and training. Basically, they were involved in most of what was required to get the DMC up and running. It was a highly collaborative effort, based on critical analysis of business needs and development of new tools and workflows to efficiently manage, move and store files, to transform metadata and to act as glueware for system integration.
Specialist knowledge and experience was key to the success of the build. Then, relevant experience in this space was scarce, and what was available was relatively inexperienced. Most vendors and manufacturers offered solutions and products that relied heavily on resources and did not effectively support flexible integration, metadata driven workflows or serious automation. If a system didn’t do what was required, other bits were often bolted on to achieve the needed functionality. But the DMC needed something more reliable, and hopefully more scalable and longer lasting. When the DMC was completed, those ‘tools’ remained, with some continuing to operate today.
Based on the DMC, BPI was approached by BT to design and implement a novel approach to asset management as opposed to simply repurposing outdated methods, much of which the industry still struggled with. BT wanted a system which delivered well-defined system architecture layers, more flexibility, automated workflows, integration to external functions and components, and a range of more useful tools to manage file-based delivery of assets and associated components as well as any type of supporting file or document. BPI, which had evolved into ‘BPI Improve’, was the one company that had the depth of experience required. This resulted in Content Store and was tailored specifically for ITV.
Quietly but steadily, the team was becoming known in the industry. That team, now Three Media after acquiring BPI Improve in 2009, knew the secret of finding, and resolving, ‘hidden’ pain points and bottlenecks in media workflows and processes of ever-increasing complexity.
We have known for a long time that the industry was changing dramatically in terms of metadata, both in the volumes and the increasing challenges in identifying and efficiently managing it. Costs to exploit content to non-linear platforms had to fall dramatically in order to efficiently monetise their content. Three Media developed XEN:Optima specifically to achieve just that. We also knew that existing workflow modelling tools, including our own XEN:Sim at the time, did not efficiently scale to support future needs. We had to rethink and retool building on the foundations of our current product set, leading to the development of ‘XEN:Pipeline’.
Briefly, XEN:Pipeline is a highly automated metadata and content management platform that enables advanced metadata discovery and management of assets and files located across multiple unrelated storage locations and supports packaging and distribution of large volumes of content to linear and non-linear distribution points. It is supported by a flexible hierarchical data schema presenting assets and files within a single, hierarchy view, via an intuitive UI.
XEN:Pipeline incorporates AI and ML to identify and match new or unknown content with related objects within the business hierarchy and to simulate and optimise workflows ensuring content is efficiently managed and delivered, compliant with SLA requirements.
Over the years we have received very positive feedback on XEN:Sim for its workflow simulation and optimisation capabilities, but it was a tough concept to sell. We could show how to optimise workflows dynamically, but it was difficult for those who knew little about performance modelling to grasp. Today, much of the industry is catching up fast in terms of gaining a better understanding of just how critical AI and ML have become to optimisation and efficiency, but it remains a tough sell to management.
But we also knew that in order to add even greater value, XEN:Pipeline needed to be able to manage complex and big data sets and to run the simulation and optimisation workflows much faster and closer to near real time. It was also apparent that the optimisation had to be self-managing requiring few manual interventions or a requirement for users to understand the logic that drives it.
To achieve that, we undertook a year-long collaboration with Imperial College London, Arcitecta, and Big Oak Dynamics. The combined goal was to ensure that XEN:Pipeline and XEN:Sim were fast enough and flexible enough to manage big data and complex media supply chain requirements both now and in the future.
Through a very clever development programme, XEN:Pipeline now lives on Arcitecta’s Mediaflux XODB and is optimised to maximise speed and accuracy and support massive scalability. The product is designed as a Platform as a Service (PaaS), delivered via the cloud or locally, in order that it can easily support all existing and foreseeable media processing services and the optimisation of highly complex media workflows in near real time. And this in turn makes XEN:Pipeline much easier to promote and demonstrate the idea of process improvement, cost reduction, and optimisation to management – and makes for a far more productive conversation!