CONGA: distributed congestion-aware load balancing for datacenters

@inproceedings{Alizadeh2014CONGADC,
  title={CONGA: distributed congestion-aware load balancing for datacenters},
  author={Mohammad Alizadeh and Tom Edsall and Sarang Dharmapurikar and Ramanan Vaidyanathan and Kevin Chu and Andy Fingerhut and Vinh The Lam and Francis Matus and Rong Pan and Navindra Yadav and George Varghese},
  booktitle={SIGCOMM},
  year={2014}
}
We present the design, implementation, and evaluation of CONGA, a network-based distributed congestion-aware load balancing mechanism for datacenters. CONGA exploits recent trends including the use of regular Clos topologies and overlays for network virtualization. It splits TCP flows into flowlets, estimates real-time congestion on fabric paths, and allocates flowlets to paths based on feedback from remote switches. This enables CONGA to efficiently balance load and seamlessly handle asymmetry… 
View via Publisher
eecs.umich.edu
549 Citations
Highly Influential Citations
108
Background Citations
324
Methods Citations
149
Results Citations
13

549 Citations

Citation Type

Citation Type

Correct-by-Construction Network Programming for Stateful Data-Planes
TLDR
Callback State Machines are proposed, a new high-level declarative network programming abstraction which allows operators to write correct data-plane programs against global state and offer programmers useful consistency guarantees without the need to manage how global state is replicated/updated at the individual switch level.
In-Network Intelligence Control: Toward a Self-Driving Networking Architecture
TLDR
This article designs intelligent in-network devices that can automatically adapt to network dynamics by leveraging powerful machine learning adaptive abilities, and introduces a centralized management plane to enhance the collaboration among distributed in- network devices.
Topology-Aware Load Balancing in Datacenter Networks
TLDR
A proposed solution focuses on two things: accurate timely acknowledgment of packet delivery and Current load on each link, which has been implemented using the NS3 simulator and the results have been compared with the other existing state of the art solutions.
Novel Multi-Level Dynamic Traffic Load-Balancing Protocol for Data Center
TLDR
A multi-level dynamic traffic load-balancing protocol, which uses an adaptive approach of parameter setting, and results show that the MDTLB outperforms the state-of-the-art schemes in terms of both the flow completion time and throughput in typical data center applications.
Routing in Black Box: Modularized Load Balancing for Multipath Data Center Networks
TLDR
It is proved that the new algorithm can achieve perfect load balancing in multipath networks and it is shown that thenew algorithm outperforms all other load balancing algorithms in performance.
Failure Diagnosis for Datacenter Applications
TLDR
Failure Diagnosis for Datacenter Applications helps decision-makers and researchers better understand the causes of failure and formulate strategies to correct the problem.
Wireless Data Center Networks: Advances, Challenges, and Opportunities
TLDR
The virtues and drawbacks of millimeter wave, terahertz, and optical wireless communications are compared and an in-depth discussion on advances and challenges in WDCNs is provided including physical and virtual topology design, quality of service (QoS) provisioning, flow classification, data grooming, and load balancing.
Routing Keys
TLDR
Routing Keys, a scalable routing paradigm for HPC networks that decouples intra- and inter-application flow contention, is introduced, showing that both ARK and NRK significantly improve the communication runtime by up to 2.7x.
Closer: Scalable load balancing mechanism for cloud datacenters
TLDR
Closer is proposed, a scalable load balancing mechanism for cloud datacenters that complies with the evaluation of technology including the deployment of Clos-based topologies, overlays for network virtualization, and virtual machine (VM) clusters and achieves 2 to 7 times better flow completion time at 70% network load than existing schemes that work with same hardware environment.
Control Theory meets programmable Data-Plane
TLDR
This thesis introduces a novel system which changes the forwarding distribution with the intend of stabilizing a network variable, as traffic rate, link delay or loss rate, around a target value.
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 64 REFERENCES
Dynamic Load Balancing Without Packet Reordering. SIGCOMM Comput. Commun. Rev
  • Dynamic Load Balancing Without Packet Reordering. SIGCOMM Comput. Commun. Rev
  • 2007
Bottleneck Routing Games in Communication Networks
TLDR
It is proved that, for both splittable and unsplittable bottleneck games, there is a (not necessarily unique) Nash equilibrium, and the rate of convergence to a Nash equilibrium in each game is investigated.
Dynamic load balancing without packet reordering
TLDR
Contrary to popular belief, it is shown that one can systematically split a single flow across multiple paths without causing packet reordering, and proposes FLARE, a new traffic splitting algorithm that operates on bursts of packets, carefully chosen to avoid reordering.
Virtual eXtensible Local Area Network (VXLAN): A Framework for Overlaying Virtualized Layer 2 Networks over Layer 3 Networks
This document describes Virtual eXtensible Local Area Network (VXLAN), which is used to address the need for overlay networks within virtualized data centers accommodating multiple tenants. The…
Scalable, optimal flow routing in datacenters via local link balancing
TLDR
An optimization decomposition is used to prove LocalFlow's optimality when combined with unmodified end hosts' TCP, and it is shown that since LocalFlow acts independently on each switch, it is highly scalable, adapts quickly to dynamic workloads, and admits flexible deployment strategies.
VXLAN: A Framework for Overlaying Virtualized Layer 2 Networks over Layer 3 Networks. http://tools.ietf.org/html/ draft-mahalingam-dutt-dcops-vxlan-06
  • VXLAN: A Framework for Overlaying Virtualized Layer 2 Networks over Layer 3 Networks. http://tools.ietf.org/html/ draft-mahalingam-dutt-dcops-vxlan-06
  • 2013
Network traffic characteristics of data centers in the wild
TLDR
An empirical study of the network traffic in 10 data centers belonging to three different categories, including university, enterprise campus, and cloud data centers, which includes not only data centers employed by large online service providers offering Internet-facing applications but also data centers used to host data-intensive (MapReduce style) applications.
The nature of data center traffic: measurements & analysis
TLDR
The nature of traffic in data centers is explored, designed to support the mining of massive data sets, and a petabyte of measurements over two months are obtained, from which detailed views of traffic and congestion conditions and patterns are obtained.
VL2: a scalable and flexible data center network
TLDR
VL2 is a practical network architecture that scales to support huge data centers with uniform high capacity between servers, performance isolation between services, and Ethernet layer-2 semantics, and is built on a working prototype.
NVGRE-EXT: Network Virtualization using Generic Routing Encapsulation Extensions
This document describes the usage of "Network Virtualization using Generic Routing Encapsulation (NVGRE)" Extensions (NVGRE-EXT). The focus of this document is on specifying the control plane…
...
1
2
3
4
5
...