Cloud adoption is a very emotive and often a confusing set of topics. For a risk averse community such as production, taking the first step into the cloud is undoubtedly pulse raising. Factors such as cost control immediately spring to mind, as well as risking client relationships and the liability for assets that aren’t necessarily your own. Considering we are used to very predictable timetables and airtight outcomes for deliverables using on-premises solutions. Why add more stress, and risk budget overspend by going into the cloud?
Well, not wishing to boil the ocean with a single sweeping statement or approach, cloud when introduced cautiously and carefully, can greatly increase productivity and throughput at various stages of the average production workflow. For example, restrictions for content ingest from set are usually tied to the horsepower of the one or two computers we have in our racks to process rushes into an offline edit ready state. Yet, when applying the infinite horizontal scale of parallel compute available in the cloud, ingest processing can be hugely accelerated.
As many of us already appreciate the dreaded cloud egress pretty much always features when trying to devise a workflow modification in production using cloud. Ingressing bulk rushes content from a shoot into the cloud is free. The cost comes when egressing the relevant files the edit needs back to ground. But one must consider the benefits of the cloud when we are not reliant on just the one or two machines being available for rushes ingest. Potentially ten’s if not hundreds can add a game changing effect to your production’s labour budgets and turn-around times.
For example, egressing 1 hour of DNx36 1080p25 proxy (~16GB) can cost as little as £1.75. This may sound high when you counter that with, “I’ve already got machines that can do this overnight”. But when you throw infinite compute breadth at your rushes processing, as well as then intelligently and automatically transferring them to your nearline or even online storage ready for a Dailies assembly to start, your project lead times can really be reduced.
Also stepping backwards up the workflow pipeline, the cloud can assist with the offload from set without having to rely on the usual transport drive overhead. Content is invariably written to disk directly on set. So, if the upload to cloud ingest processing is directly configured on this set storage, then a download of rapidly produced offline proxies can occur automatically. Mitigating the whole transport drive wrangling process, purchase, and even potential loss. Plus, an added benefit is that all content stored in the cloud is underpinned by SLAs more watertight than your own on-premises equipment.
Another fringe benefit may be to create a production specific archive on the most applicable type of storage for the duration of the project. This can avoid the cost centre of investing in an on-premise archive strategy using technologies such as LTO, which also dictates its own labour and engineering overhead.
As mentioned at the start of this blog, there are many topics that can be touched when considering a move to cloud. Major take-aways from this would be that any cloud resource usage can easily be aligned to productions. The previous examples have focused on how technology can accelerate stages of the workflow. Another reason that cloud is the future for production companies is that it aligns itself with the economic fluctuations that a production company faces when managing the business between commissions.
The traditional CapEx model requires you to build infrastructure to suit a project’s requirement. How does this model align to the business when its quiet or when you need to expand? The ultimate argument for cloud production is that it allows you to take advantage of the expansion and contraction. Expand your production architecture when the commission requires it and contract in the fallow period. With technology such as Teradici and SRT streaming from Haivision, an editor will have no idea that they are working on a virtual cloud environment. So, align your tech to an OpEx model.
By using simple techniques in AWS such as Cost Allocation Tagging, deployment of cloud infrastructure being conducted with fully orchestrated platforms such as Tyrell Cloud can greatly assure you of success. As all cloud consumption billing and costs incurred can be reconciled against a particular production, again further advancing towards OpEx.
Nick Soper, Tyrell Cloud Platform Owner