So here we are again, gearing up for another assault on all the senses at IBC. More than 1350 exhibitors spread across 14 halls – what should you focus on to make best use of your time in Amsterdam?
In today’s increasingly file-based media workflows, file transcode operations have assumed critical importance. Gone is the old router that joins processes together and in comes the smart, scalable, reliable, hidden transcoder. I apologize in advance if some of this blog piece sounds a bit pro-AmberFin, but when we designed our new Transcode Farm Controller, we talked to many different customer types to find out what was needed and they made some interesting observations that I will share with you. Most importantly, they said that a Transcode Farm Controller must be scalable, must provide an appropriate level of redundancy to suit their application and must enable a combination of high throughput and advanced system functionality.
Well, that sounds easy enough. They also wanted it to be operationally easy because the operational staff won’t have the training to understand the low level technology of every file format and the supervisors will be too busy to spend much time on the farm. (Note how I avoided trying to tell a joke there).
What’s the big deal about transcode farm control?
The big deal is the reliability aspect. To make the problem simple. Imagine that your MAM is sending jobs to iCR via web services and suddenly someone unplugs it from the rack. How do we build the system so that the MAM doesn’t know or care that something went wrong? We ended up with 4 major components to the system:
1. The interface – this is the web service “listener” and watch folder controller that responds to commands from the MAM, from the GUI(s), from the review stations or any other component of a facility using iCR
2. The transcode node – this is the engine that does the processing of the file(s)
3. The Farm Controller – this is the brain that decides which job goes to which iCR node at what time and with what overrides
4. Network License Manager – this is the control layer that allows an iCR node to be a transcoder and allocates the various options to the farm
Put those functions together on a server and call it “The new iCR Transcode Farm Controller” and you get a single, reliable, redundant interface to the iCR Transcode capabilities. Jobs are sent to the iCR Controller and behind the scenes the system architecture is sized to achieve the required levels of redundancy and throughput required by that particular application.
By combining the Farm Controller with AmberFin’s Network Licensing Server you satisfy another requirement we were asked for – to dynamically float cost-options across the underlying server hardware and no longer have a fixed node-to-server relationships. In fact one thing we have been able to achieve is to allow an international customer to move their transcode farm around the globe on a daily basis and “follow the moon”. In other words, by utilizing the power of their international VPN, they can move jobs around the planet and process jobs when the people are sleeping. Neat!
Strangely, we were also asked by customers if we could make the transcode farm cheaper. The combination of the Transcode Farm Controller and Network License Server adds this third dimension in terms of enhanced network functionality. It provides floating software licenses for occasional functions, such as standards conversion, captioning, Dolby, audio processing and watermarking.
Network License Server makes European debut at IBC
We know how complicated some of these systems can get, so we were asked if we could make a single desktop system look and behave like a big system, so we’ve managed to use the same technology in both a single standalone Desktop PC running as the proof of concept as well as in a network of 100 or more servers / blades / VMs. Furthermore, each iCR node contains all the software required to implement the four main functions of media ingest, file transcode, playback and quality control. The license defines the functionality of a specific node at a specific time.
It goes without saying that we were asked to be sure that a user’s full capacity is always on-line. For example, if one transcoder node encounters problems, the Farm Controller will seamlessly swap the required iCR operations to a back-up node and the Network License Server will ensure it is able to carry on. It is easy to tie together job queues and the Network License Server so that the network administrator is able to gauge how many jobs were delayed based on the licensing options available within the group. This feedback gives key capacity information to administrators, allowing them to manage both costs and capacity. I’m pretty pleased that we’ve listened to our amazing customers and made something shiny and new that will help the monetization of media and give a great ROI.