Cloud Computing
Similar to high performance computing (HPC), cloud computing provides hardware for running resource-intensive computations, but while HPC focuses on completing a complex set of tasks as quickly as possible with an extraordinarily powerful set of resources, cloud computing has been traditionally dedicated to providing hardware through virtual machines (VM) of varying specifications over a long period of time in order to support ongoing services such as web applications and databases.
Digital Research Alliance (DRA)
Canada-based faculty and librarians with DRA accounts can create a VMs at any time of the year by submitting a Rapid Access Service (RAS) request. If the RAS resource limits noted below and in more detail here are not sufficient, additional resources can be requested through the annual Resource Allocation Competition (RAC) with applications due between late September and early November. Upon approval, RAC resources are granted the following April.
The DRA Cloud provides access two different types of VMs with different use cases in mind, which are listed below.
Compute Instances
Compute VMs are ephemeral and provide a middle-ground between running batch jobs on the DRA HPC clusters and Persistent VMs. They can provide a powerful solution when using software that cannot be ran on the HPC clusters either natively or via Apptainer or when running computations that can run for days and/or weeks.
For geography-related research and instruction, these instances can be particularly useful for the following use cases:
- Spinning up a customized JupyterHub instance and/or RStudio server for a workshop or class
- Working with massive datasets in a desktop application, such as QGIS, AliceVision, or Blender, where access to a GPU can be significantly beneficial
RAS Resource Limits
- # of VMs: 20
- vCPUs: 80
- RAM: 300 GB
- Storage: 10 TB
- vGPU: 16 GB of memory on an Nvidia V100
Persistent Instances
These VMs are ideal for running continuous computations, like web crawlers, and ongoing services, like web servers. It’s important to remember that services cannot be ran on the DRA Cloud indefinitely, so ensure you have a plan for your VM’s end-of-life (EOL) along with a set of clear security procedures before creating it. The DRA requires that researchers reassess their cloud services annually in April by submitting a form that restates the goals of the project and necessary resources for continued use.
For geography-related research and instruction, these instances can be particularly useful for the following use cases:
- Hosting a PostgreSQL database with PostGIS
- Running a web service with software like GeoServer, GeoNode, Ushahidi, or QGIS server.
- Developing dynamic digital exhibits or publications with Omeka and/or Scalar.
Again, when working with persistent VMs, consider any implications that may come from preserving your work once the VM is no longer available by asking yourself whether your work can be easily exported, web archived, and/or hosted externally with minimal resources and no long-term maintenance.
RAS Resource Limits
- # of VMs: 10
- vCPUs: 25
- RAM: 50 GB
- Storage: 10 TB
UBC IT
In some cases, the DRA Cloud may not be the best option for a research project. In those cases, UBC IT can provide some budget-friendly alternatives. For minimal LAMP-stack web applications, like WordPress, Drupal, or Omeka, UBC IT provides simple shared hosting solutions, while more resource-intensive applications and computation tasks can run on EduCloud VMs. Additionally, UBC IT has begun a broker system that can enable resources to be requested from large commercial cloud providers like Amazon, Microsoft, and Google with additional support.
Commercial Cloud Computing Providers
Along with UBC IT’s Hybrid Cloud Services, resources from commercial cloud computing providers can be acquired with funding and credit opportunities via UBC Advanced Research Computing (ARC). Prior to using a commercial cloud computing provider, contact UBC ARC for consultation as they can provide guidance in selecting the best provider for your project and identifying any security concerns that may be applicable. This consultation can be crucial when navigating the massive and overwhelming number of services/options presented by commercial providers.
Amazon Web Services (AWS)
To further support researchers in using allocated credits on AWS, UBC ARC provides access to RONIN, a web application that can simplify and increase the accessibility of clouding computing services.
Microsoft Azure
Azure integrates smoothly with Esri software, making it a strong option for research that is heavily dependent on ArcGIS.