Summary
Technologies such as systems-on-chip (SoCs), distributed switching fabrics, silicon photonics, and programmable hardware are enabling high density designs with very high bandwidth and low latency communication at the scale of a rack. In the near future, we expect to see “rack-scale computers” with 1,000s of cores and terabytes of memory, connected with bandwidth and latency comparable to today’s smaller-scale NUMA servers. These architectures are being driven by the need to increase density and connectivity between servers while lowering cost and power consumption. Early examples of these platforms have already appeared on the market from manufactures such as AMD SeaMicro, HP, and Intel and similar solutions are being deployed at large-scale companies such as Facebook and Microsoft.
These new architectures challenge traditional assumptions and raise several interesting research questions. For example, should we think of these rack-scale units as shared-memory machines, as distributed systems, or both? What is the interconnect topology for a rack-scale computer? Which parts of the communication stack should be implemented in hardware versus software? What are the correct failure models and fault-tolerant designs? How should we integrate rack-scale computers into data center networks? What new tools are needed to assist system designers? How can researchers effectively prototype and test novel ideas in this space?
The goal of this workshop is to bring together researchers and practitioners spanning the hardware, networking, storage, systems, and application domains, to discuss innovative designs and implementations for these new architectures.
Workshop Chairs
- Paolo Costa, Microsoft Research
- Dushyanth Narayanan, Microsoft Research
Agenda
Workshop Program
Session 1
- Stanko Novakovic, EPFL – Using shared non-volatile memory in scale-out software
- Jana Giceva, ETH Zurich – Rack-scale Data Processing System
- Babak Falsafi, EPFL – Heterogeneous Memory & Its Impact on Rack-Scale Computing
Session 2
- Matthew Grosvenor, University of Cambridge – Atomic Broadcast for the Rack-Scale Computer
- Alysson Bessani, University of Lisboa – Do we need Rack-scale Coordination?
- John Wilkes, Google – Think outside the rack
- Muli Ben-Yehuda, Stratoscale – Market-Driven Resource Allocation in Rack-Scale Systems
Session 3
- Torsten Hoefler, ETH Zurich – Slim Fly: A Cost Effective Low-Diameter Network Topology
- Noa Zilberman, University of Cambridge – Bridging the Gap between Networking and End-Host Computing
- Sergey Legtchenko, Microsoft Research – Reconfigurable network topologies at rack scale
- Tyler Szepesi, University of Waterloo – Towards Reconfigurable Rack-Scale Networking
Poster Session
Panel Session
- Edouard Bugnion, EPFL
- Kimberly Keeton, HP Labs
- Andrew Moore, University of Cambridge
- Tim Harris, Oracle Labs
- John Wilkes, Google
- Paolo Costa, Microsoft Research (moderator)
