2016 IEEE NetSoft Conference and Workshops (NetSoft) 2016
DOI: 10.1109/netsoft.2016.7502476
|View full text |Cite
|
Sign up to set email alerts
|

On performance of OpenDaylight clustering

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1

Citation Types

0
12
0

Year Published

2018
2018
2021
2021

Publication Types

Select...
4
2
1

Relationship

0
7

Authors

Journals

citations
Cited by 21 publications
(12 citation statements)
references
References 4 publications
0
12
0
Order By: Relevance
“…Two informal measurement standards on SDN control plane benchmarking, by the IETF [23] and the ONF [24], specify cluster performance and stability tests. The performance of ODL clustering, in terms of synchronization overhead, failure detection and failover time, was analyzed in [25], while ONOS inter-controller traffic in different scenarios was measured and modelled in [26]. An ONOS report on SDN control plane performance [27] discusses distributed design solutions considered by developers, as well as the final implementation, and demonstrates the improvements compared to an older release.…”
Section: A High-availability In Distributed Sdn Implementationsmentioning
confidence: 99%
“…Two informal measurement standards on SDN control plane benchmarking, by the IETF [23] and the ONF [24], specify cluster performance and stability tests. The performance of ODL clustering, in terms of synchronization overhead, failure detection and failover time, was analyzed in [25], while ONOS inter-controller traffic in different scenarios was measured and modelled in [26]. An ONOS report on SDN control plane performance [27] discusses distributed design solutions considered by developers, as well as the final implementation, and demonstrates the improvements compared to an older release.…”
Section: A High-availability In Distributed Sdn Implementationsmentioning
confidence: 99%
“…The Akka framework, specified in [19], is used to synchronize information between the cluster of controllers. ODL clustered controllers are limited due to the vast number of control/flow packets generated among larger clusters [20].…”
Section: A Logically-centralized Control Planementioning
confidence: 99%
“…The total time taken to serve 1000 SDN-LB embedding requests in the SC deployment takes a longer time as each write and read request is serialized, and no concurrent state modifications are allowed to take place. Previous measurements of the RAFT implementation in OpenDaylight [9], [10] have proven that the overhead of read operations in a consensus-based cluster is similar to that of the write operations, since cluster-wide reads/writes are necessary to reach consensus on the latest state values. Lastly, the distribution time of AC suffers compared to the EC model, since EC processes transactions as fast as possible and does not implement the overhead of consistency adaptation.…”
Section: Ac (Local)mentioning
confidence: 99%
“…They focus on the evaluation of the performance of leader election procedure during the controller failure scenario. Suh et al [9], [10] experimentally measure the throughput and the recovery time of a RAFT-enabled cluster with up to 5 SDN controllers for the use case of flow table reconfigurations. These works do not discuss the effect of failures on the quality of decision-making in the context of SDN applications nor do they cover the aspects of RAFT scalability for high-throughput applications.…”
Section: Ac (Local)mentioning
confidence: 99%