Default workspace configuration
A workspace provides configurable levels of compute and storage that can be tailored to meet researchers' requirements. Workspaces can scale up or down by request, allowing for multiple configurations and support for use cases from structured data analysis using in-built tooling, to large-scale model training using Azure Machine Learning services.
Storage
| Default | Increase to | |
|---|---|---|
| Shared File System | 100TB | N/A |
| Blob Storage | 5PB | N/A |
Each workspace has available an allocated default 5TB shared file system and 5PB of blob storage. The shared file system can be increased to a maximum of 100TB if required. Storage costs are based on the use of what has been allocated.
Virtual Machines
| CPU | RAM | Storage | Extra | |
|---|---|---|---|---|
| Windows | 4 | 14GB | 30GB | N/A |
| Linux | 4 | 14GB | 30GB | 16GB (shared for home directory) |
Should dedicated compute be required, Virtual Machines (VM) can be provided. The default VM configuration provides 4 CPU and 14GB of RAM and 30GB storage. Furthermore, in a Linux VM, each user benefits from 16GB allocation for home directories, divided among each user.
Other VM configurations can be requested, and users can access any of the Azure virtual machine configurations that are available within the region where the DRE hub is deployed. Workspaces can have multiple VMs of varying configurations available. To save costs, VMs are configured to auto-shutdown overnight, based on your time-zone.
Tooling
| CPU | RAM | Scales to | |
|---|---|---|---|
| Kubernetes cluster | 8 | 16GB | 10 nodes |
Workspaces provide in-built tooling that provides access to R-Studio and Jupyter Notebook. Your organisation may provide further dedicated in-built tools for use in your workspace. These tools run on a dedicated Kubernetes cluster. The default node configuration is 8 CPU and 16GB of RAM. This cluster is shared across all workspaces and can scale to 10 nodes by default. The node configuration and maximum node pool size can be changed on request.
Applications
| CPU | RAM | |
|---|---|---|
| Minimum | 750m | 1GB |
| Maximum | 1500m | 10GB |
Each application running on the cluster is guaranteed 1GB RAM and 750m CPU, with an upper limit of 10GB RAM and 1500m CPU. This applies to Shiny applications, containerised applications, and the R console.
Backup
Workspaces are backed up on a rolling 14-day basis. If a workspace is deleted in error, you can contact Service Desk within this period to recover the data before it is permanently deleted.
While the contents of the workspace are backed up, this does not apply to the local drive of any virtual machines in the workspace. All your work should ideally be saved in the Files or Blobs section of your workspace, or on Git.
Simultaneous Workspace Users
To maintain performance and responsiveness, it is recommended that workspaces have no more than 15 members. Although workspaces can support more users, each member shares the same file system and compute resources. Performance depends on their tools and data flow needs. Having more users can reduce the available compute power for each member and increase the chances of file edit conflicts, making collaboration on projects more difficult.