Survey of Consistent Software-Defined Network Updates

Foerster, Klaus-Tycho; Schmid, Stefan; Vissicchio, Stefano

doi:10.1109/comst.2018.2876749

Cited by 113 publications

(91 citation statements)

References 167 publications

(306 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Yet, despite these efforts aimed at eliminating inconsistency-caused anomalies, we have not seen any notable proposals on fine-grained controller synchronization designs which are tailored for SDN applications with specific performance metrics. The urgency to fill this gap is especially pronounced when SDN technologies are discussed in a wider range of contexts, where advanced applications are developed on top of SDN-enabled 5G, smart grid, and ISP networks; all these cases require the support of new and finer-grained controller synchronization models [9]. In this regard, we approach the controller synchronization problem with the aim of developing fine-grained controller synchronization policies for enhancing given performance metrics.…”

Section: Introductionmentioning

confidence: 99%

MACS: Deep Reinforcement Learning based SDN Controller Synchronization Policy Design

Zhang

Poularakis

et al. 2019

2019 IEEE 27th International Conference on Network Protocols (ICNP)

View full text Add to dashboard Cite

In distributed software-defined networks (SDN), multiple physical SDN controllers, each managing a network domain, are implemented to balance centralised control, scalability, and reliability requirements. In such networking paradigms, controllers synchronize with each other, in attempts to maintain a logically centralised network view. Despite the presence of various design proposals for distributed SDN controller architectures, most existing works only aim at eliminating anomalies arising from the inconsistencies in different controllers' network views. However, the performance aspect of controller synchronization designs with respect to given SDN applications are generally missing. To fill this gap, we formulate the controller synchronization problem as a Markov decision process (MDP) and apply reinforcement learning techniques combined with deep neural networks (DNNs) to train a smart, scalable, and fine-grained controller synchronization policy, called the Multi-Armed Cooperative Synchronization (MACS), whose goal is to maximise the performance enhancements brought by controller synchronizations. Evaluation results confirm the DNN's exceptional ability in abstracting latent patterns in the distributed SDN environment, rendering significant superiority to MACS-based synchronization policy, which are 56% and 30% performance improvements over ONOS and greedy SDN controller synchronization heuristics.

show abstract

Section: Introductionmentioning

confidence: 99%

MACS: Deep Reinforcement Learning based SDN Controller Synchronization Policy Design

Zhang

Poularakis

et al. 2019

2019 IEEE 27th International Conference on Network Protocols (ICNP)

View full text Add to dashboard Cite

show abstract

“…This scope has many axes that are relevant, including updating techniques and performance goals/objectives. That is why there is more than one classification of works that have addressed the problem, such as the proposal of [11]. The work consists of a profound review of the consistent updating approaches in SDNs.…”

Section: Related Workmentioning

confidence: 99%

Towards Causal Consistent Updates in Software-Defined Networks

et al. 2020

View full text Add to dashboard Cite

A network paradigm called the Software-Defined Network (SDN) has recently been introduced. The idea of SDN is to separate the control logic from forwarding devices to enable a centralized control platform. However, SDN is still a distributed and asynchronous system: events can be triggered by any network entity, while messages and packets are prone to arbitrary and unpredictable transmission delays. Moreover, the absence of a global temporal reference results in a broad combinatorial range space of event order. During network updates, an out-of-order execution of events may result in a deviation from desirable consistent network update properties, leading, for example, to forwarding loops and forwarding black holes, among others. In this paper, we introduce a study of the Transient Forwarding Loop (TFL) phenomenon during SDN updates; for this, we define a formal model of the TFL based on causal dependencies that capture the conditions under which it may occur. Based on this model, we introduce an algorithm that ensures the causal dependencies of the system oriented toward TFL-free SDN updating. We formally prove that it is sufficient to ensure the causal dependencies in order to guarantee TFL-free network updates. Finally, we analytically evaluate our algorithm and discuss how it outperforms the state-of-the-art in terms of updating overhead. transmission delays) interface to establish networking. Furthermore, an out-of-order execution of events may occur since no global temporal reference is shared between network entities and message delays are arbitrary. This leads to the following problem: when updating a network while packet flows are taking routes to their destination, an out-of-order execution could give rise to non-deterministic behavior that temporarily deviates from network properties, which in turn may result in an inconsistent network update. Moreover and as a result, the network-wide view from the controller can transitorily be inconsistent with the current data plane state, which could affect the consistency of future network updates. To ensure consistent updates, depending on the network application, the network should align with some properties, such as no Transient Forwarding Loop (TFL) and no forwarding black hole, among others. The no TFL is one of the essential network properties desired by several network applications, including traffic engineering, virtual machine migration, and planned maintenance [4]. Informally, it ensures that a packet is never forwarded along a loop back during an arbitrary time interval to a forwarding device in the network where it was previously processed. To the best of our knowledge, (i) no study formally specifies under which conditions TFLs may occur in the context of SDNs. Furthermore, (ii) no solution to this problem aligns with the distributed and asynchronous nature of the SDNs. Indeed, the proposed solutions are centralized or synchronized. In fact, a centralized-based solution is associated with memory overhead: to perform updates, it makes use of the centr...

show abstract

“…Meanwhile, switch s 7 infers that link 7,4 has enough capacity to carry F B ; then, it installs forwarding state for F B and notifies s 3 . Then s 3 moves F B onto 3,7 .…”

Section: Decentralizing For Fast Updatesmentioning

confidence: 99%

Decentralized Consistent Updates in SDN

Nguyen

Chiesa

Canini

2017

Proceedings of the Symposium on SDN Research

View full text Add to dashboard Cite

We present ez-Segway, a decentralized mechanism to consistently and quickly update the network state while preventing forwarding anomalies (loops and black-holes) and avoiding link congestion. In our design, the centralized SDN controller only pre-computes information needed by the switches during the update execution. This information is distributed to the switches, which use partial knowledge and direct message passing to efficiently realize the update. This separation of concerns has the key benefit of improving update performance as the communication and computation bottlenecks at the controller are removed. Our evaluations via network emulations and large-scale simulations demonstrate the efficiency of ez-Segway, which compared to a centralized approach, improves network update times by up to 45% and 57% at the median and the 99 th percentile, respectively. A deployment of a system prototype in a real OpenFlow switch and an implementation in P4 demonstrate the feasibility and low overhead of implementing simple network update functionality within switches.

show abstract

Survey of Consistent Software-Defined Network Updates

Cited by 113 publications

References 167 publications

MACS: Deep Reinforcement Learning based SDN Controller Synchronization Policy Design

MACS: Deep Reinforcement Learning based SDN Controller Synchronization Policy Design

Towards Causal Consistent Updates in Software-Defined Networks

Decentralized Consistent Updates in SDN

Contact Info

Product

Resources

About