Learn how to innovate across deep distributed serverless-computing stack, from distributed systems and operating systems down to computer architecture.
The tutorial is co-located with SOSP 2023.
Venue: Room 4.
October 23rd, 2023.
Koblenz, Germany.
Slides can be found here. Join the tutorial's Slack space (recommended).
Please sign up beforehand to participate in the hands-on sessions. Please bring a laptop, from which you can SSH to a remote cloud VM.
Recordings from the last year's tutorial are available on its website and YouTube.
This tutorial provides introduces the vHive ecosystem, a full-stack open-source framework for serverless experimentation and innovation. The tutorial seeks to educate the community about serverless computing architecture and benchmarking methodology, and teach the researchers from the computer systems community to use vHive for their research. The tutorial includes a number of hands-on and demo sessions on analyzing performance of serverless applications and systems, instrumenting serverless infrastructure across the stack.
We will also showcase new vHive features, namely cluster-level performance analysis with In-Vitro, distributed cloud-edge deployment with Knative atop of OpenYurt, and serverless workflow programming with Apache Airflow atop of Knative.
vHive is a full-stack open-source framework for serverless cloud and cloud-edge experimentation, integrating production technologies for cluster management, virtualization, and workflow programming.
vSwarm benchmark suite contains 30+ microbenchmarks and 8 complex applications with distributed tracing and low-level profiling support.
Come early to get help in setting up access to cloud so that you can get hands-on experience with vHive
- Tutorial goals and contents
- vHive ecosystem overview and components, namely experimentation framework, vSwarm benchmarks, cluster-level studies with In-Vitro, workflow programming with Airflow, cloud-edge deployment with OpenYurt, CPU simulation with Gem5
- Deploying Knative functions atop vHive
- Configuring the isolation technology (container, Firecracker, gVisor)
- Case study: Deploying a Video Analytics application
Enjoy your coffee!
- In-Vitro sampler tool for summarizing production traces
- In-Vitro load generator to play a trace on Knative/OpenWhisk serverless deployment, collect measurements and analyze performance
Program and deploy serverless workflows with Apache Airflow
Deploy serverless applications in a geographically distributed setting.
Designed with Mobirise