Module COMP50030: Data Centre Infrastructure: Software, Hardware, Monitoring and Security

Department: Computer Science

COMP50030: Data Centre Infrastructure: Software, Hardware, Monitoring and Security

Tied To

• G5TA09: Scientific Computing and Data Analysis (AI Infrastructure Platforms)

Pre-requisites

Corequisites

Excluded Combinations

Aims

• To understand the software and hardware environment of HPC systems hosting advanced simulation and AI applications.
• To give students the tools and training to operate and maintain such systems.
• To present the theoretical concepts to classify, assess and design HPC systems for advanced simulation and AI applications.
• To provide students with the background required to understand new developments in software and hardware.
• To understand energy efficiency and energy consumption in data centres.
• To understand how codes can exploit a data centre’s resources.
• To give students a thorough understanding of how to manage the user base.
• To provide students with the background required to understand security concerns in data centres.

Content

• Software stack (Unix) of advanced simulation and AI platforms.
• Hardware principles underlying modern microarchitecture.
• Principles of network design and architecture.
• Principles of data centre administration.
• Server provisioning tools.
• Monitoring.
• Scheduling policies.
• Authentication and validation.

Learning Outcomes

• By the end of this module, students should be able to demonstrate:
• Knowledge and understanding of modern HPC hardware architectures.
• Knowledge and understanding of modern HPC software platforms.
• Knowledge and understanding of modern data centre user management including security aspects (authentication and validation).
• Knowledge and understanding of data centre monitoring informing performance optimisation and access policy design.

• By the end of this module, students should be able to demonstrate:
• Installation, maintenance and operation of modern AI and simulation platforms.
• Automation (scripting) of workflows.
• Scheduling for larger user bases and support of interactive usage.
• Hardware maintenance strategies (user downtime, rescheduling strategies).
• Creation and operation of data centre user administration systems.
• Design and operation of data centre monitoring systems.

• By the end of this module, students should be able to demonstrate:
• Capacity for independent learning about the hardware and software infrastructure underlying modern data centres.
• Ability to maintain and operate HPC clusters.
• Capacity for independent self-learning about the security issues arising in modern data centres.
• Ability to monitor and benchmark clusters and software independently.

Modes of Teaching, Learning and Assessment and how these contribute to the learning outcomes of the module

• Two-hour lectures delivered in a single term’s two sequential four-week blocks (eight weeks total), structured as one lecture on methods followed by one lecture on exercises (labs).
• The methodology taught in the first hour is immediately followed by a second hour of exercises to consolidate student knowledge and understanding of data centre principles.
• The module content is delivered through lectures/exercises and is reinforced by self-learning sessions and formative problem sheets, equipping students with the required problem-solving capability.
• Students will be required to submit formative problem sheets throughout the academic year to check their understanding as the course progresses.
• The coursework will require attendance at 4 2-hour workshops (two per block) where student will undertake a set task (e.g. setting up a cluster environment), with assessment through a presentation on what they have achieved.

Teaching Methods and Learning Hours

Summative Assessment

Formative Assessment

Formative assessment is provided by means of formative problem sheets.