EIDF provides a range of desktops delivered to your web browser using "virtual desktop infrastructure" (VDI). Choose from a range of pre-configured virtual desktop machines (VMs) ready-built with data science toolkits for a variety of uses. Access via ssh is also available. Target users for this service are data scientists and engineers looking for cloud-based virtual machines, singly or in small clusters, with optional GPU support.AvailabilityThis service is currently available for University of Edinburgh researchers. You will need to connect from the University network (EdLAN or VPN) to access this service.Please use the EIDF Application Portal to request access. You can find help and guidance on the process under "Service Access" in the technical documentation hosted on GitHub.EIDF Application Portal link.EIDF Portal documentation [external].VM profilesEIDF VMs come in a range of sizes and “profiles” which will expand over time. Each profile has a recommended minimum size, depending on the installed software. For the best range of data science tools we recommend Ubuntu Linux version 20.04 as a starting point. EIDF desktop services: VM profilesVM ProfilePre-installed softwareMin. vCPUsMin. RAM (GB)PlainBase operating system only.14Data Science Python and RData Science Workstation providing Anaconda Python, JupyterLab, R and R Studio and GNU Development Tools24GIS Python and RData Science GIS Workstation providing GIS Software, PostgreSQL, R and RStudio, Anaconda Python, R enabled in Juypter Notebook and GNU Development Tools28VM sizesWe offer VM allocations in a range of sizes for virtual CPU count, RAM and disk space. Our smallest is 1 vCPU and 4GB RAM; the largest is 64 vCPUs and 896GB RAM. You can if you wish split the vCPU/RAM allocation we give you across more than one VM: you could divide a 32 vCPU allocation into 4 x 8 vCPU machines if that was a better fit for your application needs, for instance. You can manage this yourself through the EIDF Portal.We will ask you to justify your resource requests as part of the application process, particularly for very large VM allocations. Users needing very large memory or CPU counts may find the Ultra2 large memory service is a better option (cf. Links).User disk storageVMs come with a default disk size of up to 50GB, but you can request more. We will ask you to justify your resource requests for disk space in the multi-terabyte ranges, but allocations up to several hundred TB fall within the standard service access model.Technical documentationMore detailed documentation for power users and developers.LinksUltra2 large memory service This article was published on 2023-11-24