Resource Disaggregation in High-Performance Computing

Moderator: Balazs Gerofi (RIKEN), John Shalf (Lawrence Berkeley National Laboratory (LBNL))

Panelists: Keren Bergman (Columbia University), Doug Carmean (Microsoft Corporation), Larry Dennison (NVIDIA Corporation), Mark Wade (Ayar Labs), Takashi Miyoshi (Fujitsu Ltd)

Abstract: Disaggregation is an emerging compute paradigm that splits monolithic servers into a number of consolidated, single-resource pools that communicate over a fast interconnect. This model decouples individual hardware resources, including tightly coupled ones such as processors and memory, and enables the creation of logical compute platforms with flexible and dynamic hardware configurations. The concept of disaggregation is driven by various recent trends in computing. The increasing importance of data analytics and machine learning workloads brings unprecedented need for memory capacity, which is in stark contrast to the growing imbalance of compute-to-memory capacity ratio of traditional server platforms. The proliferation of special purpose heterogeneous computing elements promotes the need for composability, while at the same time, the increasing maturity of optical interconnects elevates the prospects of distance independence in networking infrastructure. This panel intends to explore various aspects of resource disaggregation and its implications for future high-performance computing systems.

Website: https://resdis.github.io/panel/2021/sc/