Datacenter Traffic Control: Understanding Techniques and Tradeoffs

@article{Noormohammadpour2018DatacenterTC,
  title={Datacenter Traffic Control: Understanding Techniques and Tradeoffs},
  author={Mohammad Noormohammadpour and Cauligi S. Raghavendra},
  journal={IEEE Communications Surveys \& Tutorials},
  year={2018},
  volume={20},
  pages={1492-1525}
}
Datacenters provide cost-effective and flexible access to scalable compute and storage resources necessary for today’s cloud computing needs. [] Key Method To this end, custom transport protocols and traffic management techniques have been developed to improve datacenter network performance. In this tutorial paper, we review the general architecture of datacenter networks, various topologies proposed for them, their traffic properties, general traffic control challenges in datacenters and general traffic…
information-Agnostic Traffic Scheduling in Data Center Networks with Asymmetric Topologies
TLDR
A flowlet based information-agnostic traffic scheduling mechanism that can reduce the average FCT by up to 15.35% with a realistic workload, as compared to the state-of-the-art traffic scheduling methods.
Time-Space Decoupled SnF Scheduling of Bulk Transfers Across Inter-Datacenter Optical Networks
TLDR
This paper proposes a time-space decoupled (TSD) SnF scheduling method and demonstrates that the TSD method can outperform the conventional joint method, especially when the traffic load is moderate-to-high.
Joint Availability Enhancement and Traffic Optimization of Virtual Cluster Allocation in Cloud Datacenters
TLDR
A joint optimization function is proposed to measure the risk of VC and the core bandwidth usage with a global availability constraint and an evolutionary algorithm is raised to minimize the value of the constrained optimization function.
Deadline-Aware Fast One-to-Many Bulk Transfers over Inter-Datacenter Networks
TLDR
This paper proposes a centralized admission control coupled with a scheduling algorithm, named deAdline-Guaranteed transfEr (AGE), to guarantee the deadline of admitted data transfers and utilize the network capacity efficiently.
Load balance algorithm for an OPSquare datacenter network under real application traffic
TLDR
A synthetic traffic model is provided by capturing the traffic traces of the virtualized DC applications running in a DC and the lowest path buffer (LPB) algorithm is proposed and evaluated to optimize the OPSquare DCN performance.
QuickCast: Fast and Efficient Inter-Datacenter Transfers Using Forwarding Tree Cohorts
TLDR
QuickCast is presented, a first solution to the problems of using a cohort of forwarding trees over which large inter-datacenter transfers take place and can speed up the average receiver's completion time by as much as $10\times while only using $1.04\times$ more bandwidth; further, the completion time for all receivers also improves and is faster at high loads.
G-PFC: A Packet-Priority Aware PFC Scheme for the Datacenter
  • Zheng Cui, Steven Y. Rim
  • Computer Science
    2020 21st Asia-Pacific Network Operations and Management Symposium (APNOMS)
  • 2020
TLDR
G-PFC is presented, a packet-based priority aware flow control scheme to provide low queuing delay and less pause duration for the high priority traffic that compare to low priority traffic (packets).
Optimized Flow Scheduling for Low Latency Data Center Networks
  • Computer Science
  • 2020
TLDR
This work investigates a design which exploits path abundance — offered by DCN Fat-Tree topologies — to intelligently route flows across the switching fabric, depending on their priority, to help reduce the average and tail FCT of latency-constrained flows.
...
...

References

SHOWING 1-10 OF 275 REFERENCES
DCCast: Efficient Point to Multipoint Transfers Across Datacenters
TLDR
DCCast is presented, a centralized Point to Multi-Point (P2MP) algorithm that uses forwarding trees to efficiently deliver an object from a source datacenter to required destination datacenters with low computational overhead.
A Scalable Approach to Avoid Incast Problem from Application Layer
TLDR
The proposed approach, staggered flows, is able to avoid incast problem and to make the performance better in most cases and the potential of scalability is shown that this approach can keep the performance well with the increment of node quantity.
DCRoute: Speeding up Inter-Datacenter Traffic Allocation while Guaranteeing Deadlines
TLDR
DCRoute is proposed, a fast and efficient routing and traffic allocation technique that guarantees transfer completion before deadlines for admitted requests that assigns each transfer a single path to avoid packet reordering.
Practical TDMA for datacenter ethernet
TLDR
A TDMA MAC layer for commodity Ethernet hardware is designed and implemented that allows end hosts to dispense with TCP's reliability and congestion control and is found to result in higher bandwidth for MapReduce shuffle workloads, lower latency for RPC workloads in the presence of background traffic, and more efficient operation in highly dynamic and hybrid optical/electrical networks.
Better never than late: meeting deadlines in datacenter networks
TLDR
Evaluation from a 19-node, two-tier datacenter testbed shows that D3, even without any deadline information, easily outperforms TCP in terms of short flow latency and burst tolerance, and effectively doubles the peak load that thedatacenter network can support.
Inter-datacenter bulk transfers with netstitcher
TLDR
This paper designed, implemented, and validated NetStitcher, a system that employs a network of storage nodes to stitch together unutilized bandwidth, whenever and wherever it exists, and demonstrates that the solution can perform large data transfers at a much lower cost than naive end-to-end or store-and-forward schemes.
Energy proportional datacenter networks
TLDR
It is demonstrated that energy proportional datacenter communication is indeed possible and that there is a significant power advantage to having independent control of each unidirectional channel comprising a network link.
Mahout: Low-overhead datacenter traffic management using end-host-based elephant detection
TLDR
Mahout is presented, a low-overhead yet effective traffic management system that follows OpenFlow-like central controller approach for network management but augments the design with the authors' novel end host mechanism.
Achieving high utilization with software-driven WAN
TLDR
A novel technique is developed that leverages a small amount of scratch capacity on links to apply updates in a provably congestion-free manner, without making any assumptions about the order and timing of updates at individual switches.
...
...