Abstract
Big scale, high performance and fault-tolerance, low-cost and graceful expandability are pursued features in current datacenter networks (DCN). Although there have been many proposals for DCNs, most modern installations are equipped with classical folded Clos networks. Recently, regular random topologies, as the Jellyfish, have been proposed for DCNs. However, their completely unstructured nature entails serious design problems. In this paper we propose Random Folded Clos (RFC) and Hydra networks in which the interconnection between certain switches levels is made randomly. Both RFCs and Hydras preserve important properties of Clos networks that provide a straightforward deadlock-free multi-path routing. The proposed networks leverage randomness to be gracefully expandable, thereby allowing for fine grain upgrading. RFCs and Hydras are compared in the paper, in topological and cost terms, against fat-trees, orthogonal fat-trees and random regular networks. Also, experiments are carried out to simulate their performance under synthetic traffic patterns emulating common loads present in warehouse scale computers. These theoretical and empirical studies reveal the interest of these topologies, concluding that Hydra constitutes a practicable alternative to current datacenter networks since it appropriately balance all the main design requirements. Moreover, Hydras perform better than the fat-trees, their natural competitor, being able to connect the same or more computing nodes with significant lower cost and latency while exhibiting comparable throughput.
Highlights
Datacenters are becoming critical components in modern industry and society
These theoretical and empirical studies reveal the interest of these topologies, concluding that Hydra constitutes a practicable alternative to current datacenter networks since it appropriately balance all the main design requirements
This work explores an intermediate evolutive step based on the natural idea of randomizing the interconnection pattern between switch layers of folded Clos topologies; we have denoted them as random folded Clos networks (RFC) and they were announced in [10]
Summary
Datacenters are becoming critical components in modern industry and society. To be adequately supported, current Internet services and cloud computing require extremely powerful datacenters. The Jellyfish topology was recently proposed in [34] for DCN design This network is based on building a random regular graph (RRG) on top of the rack switch layer, aiming to facilitate graceful datacenter expansion. This avoids severe problems as packet deadlock and broadcast storms and allows for extremely simple shortest multi-path routing For these reasons, this work explores an intermediate evolutive step based on the natural idea of randomizing the interconnection pattern between switch layers of folded Clos topologies; we have denoted them as random folded Clos networks (RFC) and they were announced in [10]. This work explores a more conservative evolution of RFCs, denoted as Hydra, in which the random interconnection between some switches layers are substituted by traditional ones, a la fat-tree As it will be shown, RFCs and Hydras constitute a compromise among cost, performance, scalability and expandability inside the class of the indirect topologies.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
More From: IEEE Transactions on Parallel and Distributed Systems
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.