Cambridge Research Cloud

The Cambridge Research Cloud service Azimuth is a web applications portal that provides on-demand science platforms running on top of a private on-premise OpenStack cloud. 

The Cambridge Research Cloud offers instant high-performance compute, storage and network resources directly through clients’ web browsers. It helps to avoid the expense and complexity of buying and managing physical servers or other data centre infrastructure, as well as managing and configuring private cloud resources.

It aims to reduce the complexity of configuring private scientific computing environments from scratch in the cloud by offering the ability to create complex science platforms via a user-friendly web interface.

It also enables research groups to deploy their own scalable platforms that fit their exact needs and requirements. Departmental IT or Research Groups are able to purchase cloud resources, on which to deploy their own research computing platforms.

This service is targeted at researchers across the University and research groups with dedicated IT practitioners. 

Typical Use 

High-performance computing, AI and big data analysis. Compute, network and storage resources are tuned to provide high-performance infrastructure for data processing and analysis. 

Available Platforms

• Linux workstation – provides a secure web access to a Linux cloud console (command line), or a full, graphical desktop environment. Allows easy access to advanced hardware configuration like GPUs and large-memory machines. 

• Jupyter Notebook – provides a web-based interactive platform that enables sharing and collaboration on data, particularly beneficial to researchers working on Artificial Intelligence (AI) and Machine Learning (ML). 

• SLURM batch computing cluster – provides a completely configured installation of a Slurm scheduler with the required head node and the specified work nodes. The private cluster can be accessed via SSH or the Open On Demand web interface.

• Kubernetes Ccluster – deploys a fully functional Kubernetes cluster, optionally with autoscaling. The customer can use the easily generated configuration files to customise and install packages on the Kubernetes cluster. 

   

Available Resources

The Cambridge Research Cloud is currently composed of Intel Cascade Lake Xeon Platinum 8276 virtual machine hypervisors, connected via a 50GbE Mellanox CX6 host network link and a 10GbE connection onto the Cambridge University Data Network.

Storage can be allocated from a large Ceph cluster and mounted directly on virtual machines as block devices.

Service Features

Each department or research group will get a dedicated tenancy (private virtual server room) where each researcher can create and manage their own platform.

All resources are subject to quota limits. Additional resources can be purchased through UIS’s self-service portal.

The system is designed for continuous running 24x7. The system is currently deployed at a single site (the West Cambridge Data Centre).

All platforms come with their own monitoring stack based on Prometheus and Grafana. This is accessible via a web browser, giving immediate insights into how your workload interacts with your compute environment.

The users of the service are responsible for backing up their own data to physically separate storage or off-site if required. Please read this chapter on how that can be achieved using tools provided with the service. 

How to get the service

If you are an internal University of Cambridge user, please contact our dedicated support team at support@hpc.cam.ac.uk