Abstract
Serverless computing has introduced scalable event-driven processing in Cloud infrastructures. However, it is not trivial for multimedia processing to benefit from the elastic capabilities featured by serverless applications. To this aim, this paper introduces the evolution of a framework to support the execution of customized runtime environments in AWS Lambda in order to accommodate workloads that do not satisfy its strict computational requirements: increased execution times and the ability to use GPU-based resources. This has been achieved through the integration of AWS Batch, a managed service to deploy virtual elastic clusters for the execution of containerized jobs. In addition, a Functions Definition Language (FDL) is introduced for the description of data-driven workflows of functions. These workflows can simultaneously leverage both AWS Lambda for the highly-scalable execution of short jobs and AWS Batch, for the execution of compute-intensive jobs that can profit from GPU-based computing. To assess the developed open-source framework, we executed a case study for efficient serverless video processing. The workflow automatically generates subtitles based on the audio and applies GPU-based object recognition to the video frames, thus simultaneously harnessing different computing services. This allows for the creation of cost-effective highly-parallel scale-to-zero serverless workflows in AWS.
Highlights
The advent of Cloud Computing introduced the ability to customize the computing infrastructure to the requirements of the applications through the use of virtualization.This resulted in the widespread adoption of Cloud computing for academic, enterprise and scientific workloads
The processing of these files is possible due to the implementation of a completely redesigned parser in the SCAR client. This language focuses on the definition of the resources for each function and allows to set them to use the aforementioned execution modes. This way, a performance preprofiling of the multiple stages of a scientific workflow determines whether a certain function should be executed (i) exclusively in AWS Lambda, because it complies with its computing limitations, (ii) exclusively in AWS Batch, because the application may require additional memory/execution time beyond the maximum available in AWS Lambda or, (iii) using the lambda-batch execution mode to accommodate disparate computing requirements
In order to demonstrate the benefits and performance of the platform, a use case has been defined that builds a serverless workflow to perform frame-level object detection in video together with the inclusion of subtitles from the audio transcript, with potential applications in surveillance. This demonstrates the ability of SCAR to provide an eventdriven service for multimedia processing, triggered by video uploads to an S3 bucket that can automatically scale up to multiple function invocations and several EC2 instances to cope with the workload and automatically scale down to zero provisioned resources
Summary
The advent of Cloud Computing introduced the ability to customize the computing infrastructure to the requirements of the applications through the use of virtualization. The main scientific challenge addressed in this contribution is to provide event-driven serverless workflows for data processing that simultaneously feature scale-to-zero, high elasticity and the support for GPU-based resources. This is achieved through the integration in SCAR of AWS Batch [5], a managed service to provision virtual clusters that can grow and shrink depending on the number of jobs to be executed, packaged as Docker containers. Definition Language that can simultaneously use both Lambda and Batch for the execution of data-driven applications composed of multiple steps This results in a tool that can foster serverless computing adoption for multiple enterprise and scientific domains, supporting any CLI-based file-processing application packaged as a container image.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
