Instructor: Grigory Shamov (University of Manitoba)
Title: Using Clouds and Containers
A beginner’s introduction to using two kinds of technologies: Clouds and Containers. The morning session will cover Clouds/VM technology; the afternoon session will build on that, explainng a related technology, Linux Containers, of which Singularity is one of the most popular ones for the HPC workloads.
Objectives
- Give a general intro to the Virtual Machines and Linux Containers
- Getting started using ComputeCanada Arbutus OpenStack Cloud
- Getting started with Singularity HPC containers
Target audience: everyone interested in using Clouds and/or HPC
Level: beginner (Morning) to intermediate (Afternoon)
Prerequisites: basic knowledge of Linux command line; basic familiarity with installing software on one of the Linux distributions (Ubuntu, CentOS, etc.).
Setup:
- Laptop software: SSH client (built-in on Mac/Linux, on Windows MobaXTerm or PuTTY)
Morning session: Using ComputeCanada OpenStack Cloud
Outline:
- What is Cloud? Cloud services delivery modes, private and public clouds.
- Virtualization technology: Virtual Machines, and contaners.
- Overview of the ComputeCanada Arbutus system.
- Walkthrough of creating a VM on Arbutus.
- Securing your VMs, installing software, etc.
- Next steps: ways to organize your computations on the Cloud.
Duration: 3 hours
Afternoon session: Containers for HPC: Singularity
Outline:
- What are the Linux Containers, and how are they different from the VMs?
- What are they good for? Reproducible computing, and Mobility of computing.
- A brief overwief of Docker and DockerHUb
- From Docker to Singularity
- Installing Singularity (might use cloud VMs created in the Morning session)
- Building and Exploring the containers: ins and outs, accessing filesystems
- Using Docker containers with Singularity
- Intro to Singularity build service and container repository: SingularityHub
- Next steps: examples of production-ready use cases for Singularity (OSG, TensorFlow, BioIformatics pipelines, custom build environments).
Duration: 3 hours