Table of Contents
Proof of Concept: Kubernetes@ICS
See the kubernetes@ICS page. Send a message to firstname.lastname@example.org if you would like to trial our kubernetes cluster.
Instructional Virtual Host Environment
ICS provides a computer cluster that is dedicated to courses that can benefit from using virtual hosts. Instructors are encouraged to sign up or send question by sending email to email@example.com.
Running a VM:
General Instructions for running a vagrant virtual machine on ICS hosts.
TL;DR Run on any openlab host:
Courses Specific VM's:
So far we have setup Vagrant virtual environment for the following courses.
- Resource distribution, ensuring that student work utilizes cycles across all nodes.
- Restricting resources to enrolled students.
- Limiting the resources that a single student can utilize.
- Garbage collection or freeing resources when a student is done.
- Each student receives a virtual machine.
- VM's are started by running a single command from command line.
- Resource distribution are managed by ICS Grid Engine
- Grid Engine transparently routes job to host with most free resources. Advanced users can choose to exercise more control over the process.
- Two ways to work: Student may work from the console of the VM or connect remotely via ssh.
- Suspend and resume VM's.
- Using programs like `screen` or `tmux`, students can run VM's indefinitely.
- Orphaned VM's are identified and removed frequently.
Local Virtual Machines
Virtual box and Vagrant are available on most ICS instructional and research compute hosts. Any user with login privilege may create a virtual machine using these utilities. Please coordinate long term virtual machine deployment with computing support by sending mail to firstname.lastname@example.org.
Cloud computing provides a cost effective way for researchers to extend their computing without significantly increasing their capital expenditures. Computing support can project ICS services into the Google Cloud Platform. Computing that is not used, turned off, is not typically charged but can be scaled up quickly when additional resources are needed.