Sharing the Data Center Network

By Alan Shieh, Srikanth Kandula, Albert G. Greenberg, Changhoon Kim, Bikas Saha. Symposium on Networked Systems Design and Implementation, 2011

409 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/a91802791a5d8c93fc3fb3843286aedf741e33cd

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Performance Isolation and Fairness for Multi-Tenant Cloud Storage (2012, unknown): “While more throughput efficient, Gatekeeper’s ingress and egress scheduling [27], Seawall’s congestion-controlled VM tunneling [28], and FairCloud’s per-endpoint sharing [26], respectively provide fairness on a per-VM, VM-pair, or communicating-VMgroup basis, rather than on a per-tenant basis.”
Multi-resource packing for cluster schedulers (2014, background): “Some prior work focuses on fair network allocation even when the applications are uncooperative [22], as well as better ways to schedule network transfers [10] and to steer around network hotspots [11].”
Planck (2014, background): “Similarly, Seawall [35] and EyeQ [21] monitor and shape traffic at the edge, but do not monitor switches.”
The only constant is change: incorporating time-varying network reservations in data centers (2012, background): “Seawall [34] and Netshare [26] propose bandwidth slicing mechanisms that aim to provide fair sharing of networks with minimum bandwidth guarantee and statistical multiplexing, but do not provide deterministic bandwidth guarantees.”
Achieving high utilization with software-driven WAN (2013, background): “Sea-wall [31] uses TCP-like adaptation to converge to fair share, but high RTTs on the WAN would mean slow convergence.”
DeTail: reducing the flow completion time tail in datacenter networks (2012, background): “Datacenters are typically characterized by a shared environment with many applications vying for resources [31].”
Resource Management in Clouds: Survey and Research Challenges (2015, background): “[195] propose Seawall, a system in which bandwidth is shared amongst contending VMs to achieve max–min fairness via the tunneling of traffic through congestion controlled, point-to-multipoint tunnels.”
Efficient coflow scheduling with Varys (2014, unknown): “The impact of coﬂows on job durations and their network foot-print have received extensive attention [15, 17, 33, 35, 39].”
Decentralized task-aware scheduling for data center networks (2014, background): “Cosmos [26] 55 70% of tasks involve 30-100 flows, 2% involve more than 150 flows”
FairCloud: sharing the network in cloud computing (2012, methodology): “Second, PS-N (and we believe PS-P as well) could be implemented through a distributed mechanism similar to Seawall [25].”
Fastpass (2014, background): “In multi-tenant datacenters, Seawall [32] provides weighted fair sharing of network links, and EyeQ [22] enforces trafﬁc constraints.”
Data Center Network Virtualization: A Survey (2013, methodology): “As a result, the network is vulnerable to insider attacks such as performance interference and Denial of Service (DoS) attacks [6].”
End-to-end Performance Isolation Through Virtual Datacenters (2014, background): “These services are often implemented using appliances, which include in-network middleboxes like load balancers and end-devices like networked storage servers.”
this paper (2009, unknown): “Streaming like this significantly improves users’ experience and enables new applications.”