Dynamic load balancing without packet reordering
By Srikanth Kandula, Dina Katabi, S. Sinha, Arthur W. Berger. CCRV, 2007
355 citations recorded by the source corpus.
Subfields: Computer Networks and Communications, Electrical and Electronic Engineering
View Paper: https://www.semanticscholar.org/paper/7246dadb0a1eacd22232bf0d283b22d1b59c4546
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Networks and Communications
- Changhoon Kim
- Mohammad Alizadeh
- Jennifer Rexford
- George Varghese
- Brighten Godfrey
- Srikanth Kandula
- Yashar Ganjali
- Anirudh Sivaraman
In Electrical and Electronic Engineering
- George Varghese
- Jennifer Rexford
- Brighten Godfrey
- Ian F. Akyildiz
- Yashar Ganjali
- Srikanth Kandula
- Pu Wang
- Albert G. Greenberg
Citations in the corpus
- HULA: Scalable Load Balancing Using Programmable Data Planes (2016, background): “In order to avoid packet reordering, HULA load balances at the granularity of flowlets [11]— bursts of packets separated by a significant time interval.”
- A roadmap for traffic engineering in SDN-OpenFlow networks (2014, background): “Although these schemes provide a good performance when operating with static load balancing, they are unsuitable for the dynamic load balancing protocols [22], since this static mapping of flows to paths does not account for either current network utilization or flow size, which results in collision”
- DRILL: Micro Load Balancing for Low-latency Data Center Networks (2017, background): “category include balancing “flowlets” [19, 44] or per-packet spreading of traffic in a round robin fashion [27, 34].”
- CONGA: distributed congestion-aware load balancing for datacenters (2015, methodology): “This example illustrates a fundamental limitation of any local scheme (such as Flare [27], LocalFlow [44], and packet-spraying [10]) that strictly enforces an equal traffic split without regard for downstream congestion.”
- FlowRadar: A Better NetFlow for Data Centers (2016, neutral): “We define flowlets as bursts of packets from a flow that are separated by gaps larger than a time slot [27].”
- The nature of data center traffic: measurements & analysis (2009, methodology): “When explicit begins and ends of a �ow are not available, similar to much prior work [26, 30 ], we use a long inactivity timeout (default 60s) to determine when a �ow ends (or a new one begins).”
- Achieving high utilization with software-driven WAN (2013, neutral): “, FLARE [18]) or host-level mechanisms (e.”