AWS for M&E Blog
How AWS customer Motive optimized video cost using AWS Elemental MediaConvert
by Lokesh Angadi Pethaiah, Andy Robinson, and Prashant Tyagi
This post was written in collaboration with Inderjeet Singh, vice president engineering, and Chandra Rathina, manager engineering, from the Motive team
Motive is a company that has developed a tracking and fleet management solution for trucking companies. As Motive developed its product offerings, it realized that its solution applies not only to the trucking industry but also to many businesses and organizations that operate in the physical economy, such as construction, agriculture, field service, manufacturing, energy, and many more. The common thread that runs through all these industries is the need to connect and automate their operations. Without Amazon Web Services (AWS) Cloud adoption, these businesses cannot efficiently grow to meet the ever-increasing demand for their output.
Growth and optimization in AWS
Motive saw a 21 percent increase in spending with AWS due to the spike in demand in the trucking and logistics industry during the peak of the COVID-19 pandemic. In particular, its usage of Amazon Elastic Transcoder, a file-based video transcoding service, was one of the top three cost contributors to the company’s spend. In this blog, we examine how Motive uses Amazon Elastic Transcoder for its video workflow and how we worked together to save 85 percent of the cost by migrating this video workload to AWS Elemental MediaConvert (MediaConvert), a file-based video transcoding service with broadcast-grade features.
Motive’s original video workflow
Motive customers deploy electronic logging devices (ELD) and dashcams on their trucks to provide tracking, fleet management, and safety monitoring for drivers. The dashcams operate by uploading locally generated clipped videos in a TAR bundle to a bucket in Amazon Simple Storage Service (Amazon S3), an object storage service. This upload starts an event in AWS Lambda, a serverless, event-driven compute service, through an event initiation that extracts the bundle and writes the files back to the bucket. Motive built a mechanism to submit jobs to transcode the files, track the jobs, perform callbacks, and store the status upon successful completion of the jobs. A job is submitted for each file that produces MP4 and WebM videos. These videos are made available to view for users, analyzing driver safety and rating them for their driving skills. The diagram below depicts the architecture flow of the transcoding jobs.
Figure 1. Motive video transcoding solution architecture (original)
On average, Motive processed tens of thousands of videos per day but could soon be scaling up in those numbers. During discovery and analysis of the usage pattern, Motive used mainly the MP4 videos and not WebM videos. The customer is charged for use of Amazon Elastic Transcoder per job, and the minimum billable duration is 1 minute. Assuming the video duration of the transcoding job is less than a minute, we round up to 1 minute for billing.
AWS Cost Explorer has an easy-to-use interface that lets you visualize, understand, and manage AWS costs and usage over time. The AWS Cost Explorer chart below shows the pricing details of the Amazon Elastic Transcoder. The weekday charges are around US$3,000, and the weekend charges are less due to less trucking traffic.
Chart 1. AWS Cost Explorer output of Motive’s Amazon Elastic Transcoder charges
Upon learning Motive’s video transcoding requirements, we recommended MediaConvert for its transcoding jobs. MediaConvert provides a comprehensive suite of file-based video transcoding features that address the needs of most use cases. MediaConvert is optimized to improve scalability, which allows more files to be processed in parallel. The picture below depicts the modified architecture flow of transcoding jobs.
Figure 2. Motive video transcoding architecture (revised)
With MediaConvert, customers can choose from on-demand pricing, reserved pricing, or a combination of the two, depending on workload and video transcoding requirements. We recommended Motive start with on-demand pricing. With on-demand pricing, the duration of each output is calculated in 1-second increments and then converted into fractional seconds to determine the total charge for the output. For example, an output with a duration of 45 seconds would be 45/60, or 0.75 of a minute. The minimum duration for billing is 10 seconds, so an output with a duration of fewer than 10 seconds will be billed for 10 seconds. This leads to substantial savings for customers like Motive. The AWS Cost Explorer chart shows the current charges in MediaConvert. Today, the per-day charges are around US$300.
Chart 2. AWS Cost explorer output of Motive AWS Elemental MediaConvert charges
The AWS Cost Explorer chart below shows how the Motive transcoding charges were reduced after migrating the workload to MediaConvert.
Chart 3. AWS Cost Explorer comparison of Amazon Elastic Transcoder and AWS Elemental MediaConvert
AWS Elemental MediaConvert vs. Amazon Elastic Transcoder
In this section, we share the test results of customer video files to provide more insights into the compression and use of QVBR. The artificial intelligence (AI) smart cameras record driver-facing and road-facing videos with different resolutions and file characteristics. The table below shows the sample videos with the original file size and transcoded file size to showcase the compression savings achieved when switching from Amazon Elastic Transcoder to MediaConvert.
Transcoded File Size Comparison
| A | B | C | D | E | F | |
| 1 | Elastic Transcoder Service | MediaConvert | ||||
| 2 | Customer Video Files | Duration | Original File Size (MB) | MP4 Fe Size (MB) | MP4 File Size (MB) | Compression Saved with MediaConvert |
| 3 | driver_facing-dc34.mp4 | 24 | 2.85 | 3.6 | 1.8 | 36.84% |
| 4 | front_facing-dc54.mp4 | 30 | 17.9 | 33.4 | 11.1 | 37.99% |
| 5 | front_facing-dc34.mp4 | 22 | 8.52 | 19.8 | 5.71 | 32.98% |
Video transcoding savings
MediaConvert provides additional cost savings when handling short-duration files because billing is charged for a minimum of 10 seconds, whereas Amazon Elastic Transcoder has a minimum billing duration of 60 seconds. For Motive, this was a significant savings because its typical video files were around 30 seconds.
| A | B | C | D | E | F | G | |
| 1 | Elastic Transcoder Service | MediaConvert | |||||
| 2 | Customer Video Files | Duration | Billable Duration(s) | Cost | Billable Duration(s) | Cost | MediaConvert Cost Savings |
| 3 | driver_facing-dc34.mp4 | 24 | 60 | $0.030 | 24 | $0.006 | 80.00% |
| 4 | front_facing-dc54.mp4 | 30 | 60 | $0.030 | 30 | $0.008 | 75.00% |
| 5 | front_facing-dc34.mp4 | 22 | 60 | $0.030 | 22 | $0.006 | 81.67% |
Reduction in Amazon S3 storage and content delivery
When using QVBR video compression in MediaConvert, the output video file size is also reduced, typically over 30 percent compared to Motive’s previous workflow. Smaller files mean that Motive’s storage costs in Amazon S3 are also reduced, as well as its content delivery network (CDN) costs for the transcoded videos using Amazon CloudFront, a CDN service built for high performance, security, and developer convenience.
Summary
As Motive grows to support more customers and industries, the number of daily ingested videos might reach hundreds of thousands, which could have a significant impact on transcoding charges. With the help of AWS-native services, like AWS Cost Explorer and AWS cost and usage reports, customers can monitor the costs in near real time and customize the reports based on the requirements and services that they would like to monitor. With this video migration to MediaConvert, Motive can easily continue to scale to 10 times its current video workflow pricing and still won’t exceed the current transcoding charges. Along with adding new services and new features to the existing workflow, we ran a cost-optimization workshop to understand the customer usage and provided a recommendation that helped Motive to reduce its AWS billing charges.