The amounts of data generated by user applications exceed the processing capacity of any single computer. Networked systems are built from a multitude of computing components, connected through the network.
A prominent example of networked systems are data centres, which host tens of thousands to hundreds of thousands of computing nodes. Networked systems are often driven by performance, targeting metrics such as throughput, latency and power efficiency.
Our group's research explores questions such as:
- How to increase the scalability of data centre networks?
- Can we increase power efficiency within the data centre?
- What are the latency components within the data centre and how can we reduce them?
This research integrates performance measurements with hardware design and application-level algorithms, and requires extensive knowledge across a breadth of topics.