Transcoding in the Cloud

Fourth of July question: Shoot for the clouds or keep transcoding on the ground? July 11, 2017
XXX

I watched some amazing fireworks this past Fourth of July weekend, with powerful rockets screaming skyward to burst among the clouds and fantastic fountains erupting brilliantly at ground level. After the excitement, I found myself musing on the show’s parallel to the question of whether it is better to transcode in the cloud or to use on-premise servers.  People want to understand how cloud-based transcoding might benefit their business. Do we even need on-premise transcoding anymore?

The truth is that you can’t discuss transcoding without considering the workflow around it. Let’s start with two basic definitions:

1.      Transcoding is the adaption of a media file from one format to another: converting a 1080i file to 720p or Avid DNxHD to H.264 web-compatible format, changing the bit rate, or making numerous other changes needed to deliver content to a particular platform.

2.      Workflow is the path content takes between creation and delivery. A workflow can include manual processes such as Standards & Practices, machine-automated processes such as quality control, and, yes, transcoding.

In other words, transcoding is just one activity within the overall journey we call a “workflow.” What happens upstream and downstream of transcoding matters.  We are really asking about content operations, which come in two flavors: Production Content that is in the process of being created and Distribution Content that is relatively complete but needs to be adapted for one or more distribution platforms.

Production content is the “raw material” of the production factory, from a YouTube producer creating a workout video to a broadcast newsroom creating this evening’s newscast.  Transcoding may be needed to bring in external content (such as from a news-wire service, archive or external library) or to share content between departments running different systems (such as production and promotions departments).  Production content workflows tend to be heavily time-dependent, frequently operate on smaller files and typically require very few destination formats and variations.

Distribution content is the finished goods that need to be packaged for delivery.  The destination could be a network’s master control, iTunes, Roku, SVOD or any other distribution path.  It tends to be longer in duration, although it may also include commercials and promos, and generally requires conversion into significantly more formats and varieties.

With all that in mind, let’s re-ask the question: cloud-based or on-premise transcoding? The answer often comes down to three simple factors:

TIME

Transcoding large files can take significant time; uploading large files to the cloud, potentially even longer.  If the delivery need is measured in minutes, as is frequently required with production operations, it’s unlikely cloud services will reliably meet that demand.  If the need is measured in days, which is normal in distribution operations, the cloud becomes a more realistic option.

Internet connection speed also greatly affects time.  Just comparing one recent Positive Flux client with a 100Gig internet backbone and another with only 200Mb, the client with the faster pipe was far better positioned to utilize cloud services for a wider variety of services. 

DISTANCE

A widely-dispersed team whose members need to operate on the content at the same time could benefit from a cloud-based solution, if it has the workflow tools they need.  For them, the “time cost” of bringing content to the cloud is offset by the benefits in freedom of location.

If the content is destined for multiple off-site locations, then time spent to upload it into the cloud is not wasted. In fact, it has the effect of moving the content closer to all its destinations and potentially reducing the number of uploads to just one. We can then leverage cloud-based workflows and transcoding to deliver the many variations to the different destinations. 

If the content is only destined for an in-house master control facility, then cloud-based operations make less sense. Why go to the cloud only to return to the same starting point?

MONEY

Moving transcoding and workflow operations to the cloud trades capital investment for operational cost.  Cloud-based operations could help avoid buying some servers and software licenses, or building a data room or complex networks, but careful financial modeling and good estimates of content volumes are critical.  The pay-as-you-go model might help with cash-flow, but those with higher volumes may find themselves penalized. Cost calculations need to include upload, storage, transcode operations, and other processing and delivery charges from the cloud service provider.  It is also important to amortize the incremental internet service used to and manage content operations.

On-premise solutions provide an all-you-can-eat solution with fixed pricing that should be easier to budget. Licenses can be used for years, although they incur annual support fees.  Right-sizing is critical here: buy too many licenses and engines sit idle; buy too few and content piles up like flights into O’Hare. In our practice, 4,000 one-hour transcodes per year is the rule-of-thumb for the economic break-even point for one server.

People ask if there are maintenance or IT cost savings associated with using cloud services. It seems like this should be the case, but the reality is that web services need to be managed and monitored just like local servers. Companies find themselves needing the same size team either way.

WHY NOT HAVE IT ALL?

For many workflows, the decision of where to perform transcoding is not an A/B choice. One of the cloud’s many advantages is its elasticity.  It’s easy to spin up additional transcode engines in the cloud when a big job comes through or when many projects need to be accomplished at the same time.  Although on-premises solutions do not offer this capability today, it’s conceivable that on-premise technology providers could soon provide the same elasticity, providing a hybrid on-site licensing/virtualization model or hybrid onsite/cloud service, which would manage the flow of content and scale on demand.

Like the fireworks, perhaps the best show of all ties together earth-bound launch platforms and cloud-based firepower.

Share your experiences in deciding to cloud or not to cloud media workflow operations in the comments.

Larry Thaler is the President of Positive Flux, a consulting firm that specializes in helping media companies take advantage of the rapid changes occurring in the industry.  He can be reached via TV Technology.

Receive regular news and technology updates.
Sign up for our free newsletter here.

Comments

Twitter