TY - GEN
T1 - ExRec
T2 - 16th ACM/IEEE Symposium on Architectures for Networking and Communications Systems, ANCS 2021
AU - Zerwas, Johannes
AU - Avin, Chen
AU - Schmid, Stefan
AU - Blenk, Andreas
N1 - Publisher Copyright:
© 2021 ACM.
PY - 2021/12/13
Y1 - 2021/12/13
N2 - In order to meet the increasingly stringent throughput and latency requirements in datacenter networks, several innovative network architectures based on reconfigurable optical topologies have been proposed. Examples include demand-oblivious reconfigurable topologies such as RotorNet (SIGCOMM 2017), Opera (NSDI 2020), and Sirius (SIGCOMM 2021), as well as demand-aware topologies such as ProjecToR (SIGCOMM 2016). All these architectures feature attractive performance properties using specific prototypes. However, reproducing these experiments is often difficult due to missing hardware and publicly available software. This paper presents a flexible framework for reconfigurable networks based on off-the-shelf hardware, which supports experimentation and reproducibility at a small scale. We describe how our framework, ExReC, can be instantiated with different configurations, allowing us to emulate existing architectures and to study their trade-offs. Finally, we demonstrate the application of our approach to different use cases and workloads, including distributed machine learning training.
AB - In order to meet the increasingly stringent throughput and latency requirements in datacenter networks, several innovative network architectures based on reconfigurable optical topologies have been proposed. Examples include demand-oblivious reconfigurable topologies such as RotorNet (SIGCOMM 2017), Opera (NSDI 2020), and Sirius (SIGCOMM 2021), as well as demand-aware topologies such as ProjecToR (SIGCOMM 2016). All these architectures feature attractive performance properties using specific prototypes. However, reproducing these experiments is often difficult due to missing hardware and publicly available software. This paper presents a flexible framework for reconfigurable networks based on off-the-shelf hardware, which supports experimentation and reproducibility at a small scale. We describe how our framework, ExReC, can be instantiated with different configurations, allowing us to emulate existing architectures and to study their trade-offs. Finally, we demonstrate the application of our approach to different use cases and workloads, including distributed machine learning training.
KW - data centers
KW - demand-aware
KW - reconfigurable networks
UR - http://www.scopus.com/inward/record.url?scp=85124156448&partnerID=8YFLogxK
U2 - 10.1145/3493425.3502748
DO - 10.1145/3493425.3502748
M3 - Conference contribution
T3 - ANCS 2021 - Proceedings of the 2021 Symposium on Architectures for Networking and Communications Systems
SP - 66
EP - 72
BT - ANCS 2021 - Proceedings of the 2021 Symposium on Architectures for Networking and Communications Systems
PB - Association for Computing Machinery, Inc
Y2 - 13 December 2021 through 16 December 2021
ER -