Cloud Encoding vs. An Encoder in the Cloud

Encoding.com pioneered cloud encoding as a service over 6 years ago. In 2008 the transcoding options for high volume media companies were limited; there were transcoding hardware appliances, open source software, or enterprise encoding software. All of these options were expensive, difficult to scale, and required on going maintenance. Over the past few years many of the enterprise hardware and software providers have launched cloud encoding offerings, Elemental, Harmonic, Sorenson, & Telestream, to name a few. At first glance these new offering seem competitive to Encoding.com, they all have cloud in the title and offer video encoding so what's the difference? All of these services lack several important features that make operating them at scale difficult, costly, and inefficient. In the interest of saving many hours of research, testing, and wasted cloud compute hours we have outlined the differences between an encoder running in the cloud and cloud video encoding:

Encoder in the Cloud: Encoding software is made available through a cloud compute provider. For example Elemental, Sorenson, Harmonic, and Telestream are all available via the Amazon Marketplace where you can login to your AWS control panel, start the EC2 instance of your choice with a preconfigured version of the software installed. Your AWS account is then charged a monthly subscription access fee as well as an hourly for the EC2 instance plus an additional hourly fee charged by the software vendor. Prices range from $1.00 $8.00 per hour depending on the vendor and size of the EC2 instance chosen.

Cloud Encoding: The notion of compute and software configuration are completely removed from the equation to users of true cloud encoding vendors. The operator integrates with one of the available cloud vendors integration methods, API, Watch Folder, Desktop App and can send highly variable volume to the provider and be billed only on usage (GBs or minutes).

Feature Encoder in the Cloud True Cloud Encoder

Auto Compute Scale If encoding volume reaches more than the capacity of one of the cloud encoding servers can handle, manual operator intervention is required to login to the AWS interface and start another one. The cloud provider will intelligently scale up compute resources based on demand and some offer SLAs around queue time, or simultaneous job requests.

Auto Compute Scale Down When encoding volume dips below the capacity of the cloud encoding servers currently running manual operator intervention is required to login to the AWS interface and turn off encoding resources. At hourly rates of up to $8.00 scaling down can be just as important as scaling up and thousands of dollars worth of compute are wasted on empty compute hours in a high volume workflow. Since cloud encoding providers charge by the minute, being able to spin down compute instances in real-time with changes in workflow is critical to cost savings and efficiency

Intelligent Job Management With a workflow that requires more than one encoder in the cloud these system lack any centralized job management and force the operator to develop an in house encoding queue to distribute job volume evenly to the currently running servers. For example if a job gets stuck or an error occurs no automated facility exists to restart the job or gracefully redistribute other jobs in the queue. Job volume is distributed to compute resources automatically and in processed in parallel. If 30 jobs are sent simultaneously each with 10 outputs, each of the 300 outputs are processed in parallel in the same amount of total job time required to 1 job with 10 outputs.

System Encoding Analytics Since each of the encoders in the cloud operate as separate workers, no central analytics or encoding queue is present to inform the operators of overall encoding volume, queue times, processing times, error reporting, etc. Centralized web based job configuration, encoding queue and analytics are provided to manage a large encoding volume. These realtime tools allows operators complete transparency and visibility into critical system metric over time.

We believe the encoder in the cloud model would be a stretch at best to label as cloud computing and definitely does not qualify as a software as a service. This model would be analogous to salesforce.com requiring you to manually start additional servers (paid by the hour) as your use of their system increased!

Harmonic, Inc. recognized the difference between an encoder in the cloud and a cloud encoder and even though they maintain their ProMedia Carbon in the Amazon Marketplace have partnered with Encoding.com to make ProMedia Carbon available in a true cloud encoding model. Now users can take advantage of the tremendous video quality of the ProMedia Carbon toolset while taking advantage of the elastic capacity and advanced toolset of a true software as a service delivery model.

LINK:	http://www.encoding.com/blog/2014/04/23/real-cloud-encoders-vs-fake-cl...
	See more stories from encoding

Most recent headlines