A multi-stage graph based algorithm for survivable Service Function Chain orchestration with backup resource sharing

Kibalya, Godfrey; Serrat, Joan; Gorricho, Juan‐Luis; Serugunda, Jonathan; Zhang, Peiying

doi:10.1016/j.comcom.2021.04.008

Cited by 12 publications

(4 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Content may change prior to final publication. For the intra-domain evaluation and mapping of the assigned sub-SFCs, an InP can use any single-substrate SFC provisioning algorithm such as [10]- [14] depending on its intra-domain policies. Moreover, under a multi-domain setting, it is possible that different InPs run different intradomain algorithms for service provisioning.…”

Section: Consensus and Binding Stepmentioning

confidence: 99%

“…Moreover, under a multi-domain setting, it is possible that different InPs run different intradomain algorithms for service provisioning. In this paper, we adopted the algorithm proposed in [10] for SFC provisioning with some minor modifications to suit the sub-SFC mapping. Using the intra-domain candidate nodes and links for the assigned sub-SFC, the algorithm constructs a multi-stage graph similar to the one in Fig.…”

Section: Consensus and Binding Stepmentioning

confidence: 99%

“…The feasible nodes and links associated with the least cost between the fictitious ingress and egress nodes, considering both resource and energy consumption, are chosen for hosting the assigned sub-SFC. We refer the interested reader in the details of the algorithm to consult reference [10].…”

Section: Consensus and Binding Stepmentioning

confidence: 99%

“…In this regard, end-to-end services will be instantiated as service chains consisting of an ordered set of Virtual Network Functions (VNFs), commonly refereed to as Service Function Chains (SFCs), which can be easily and dynamically deployed, scaled or migrated [7]- [9]. The problem of service deployment considering a single administrative network infrastructure has been extensively addressed in the literature [10]- [14]. However, for VOLUME 4, 2016 a number of practical scenarios such as Internet-of-Things applications, the geographical location at which the data is generated may be different from where it will be accessed, processed or consumed [15], [16].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

A Multi-Stage Graph Aided Algorithm for Distributed Service Function Chain Provisioning Across Multiple Domains

et al. 2021

Self Cite

View full text Add to dashboard Cite

Network Service Providers (NSPs) envisage to support the divergent and stringent requirements of future services by instantiating these services as service chains, commonly referred to as Service Function Chains (SFCs), that are customized and configured to meet specific service requirements. However, due to the limited footprint of the Infrastructure Providers (InPs), these SFCs may have to transcend multiple InPs/domains. In this regard, determining the optimal set of InPs in which to embed the SFC request emerges as a complex problem for several reasons. First, the large number of possible combinations for selecting the InPs to embed the different sub-chains of the request makes this problem computationally complex, rendering optimal solutions only after long computations, especially in large scale networks, which is unfeasible for delay sensitive applications. Second, the unwillingness of InPs to disclose their internal information, which may be vital for making embedding decisions, usually implies the provisioning of single-domain solutions, which are unsuitable in this working scenario. In this regard, this paper first formulates the multi-domain service deployment problem under multiple request constraints, such as bandwidth or delay, among others. Then, due to the NP-hardness nature of the above problem, this paper proposes an algorithm that is aided by a multi-stage graph for computing a request embedding solution in a distributed manner, solving the problem in acceptable run-times. Results from different simulations reveal that the proposed algorithm is optimized in terms of acceptance ratio and embedding cost, with up to 60.0% and 88.7% improvements in terms of embedding cost and execution time, respectively, for some scenarios, in comparison with a benchmark state-of-the-art algorithm.

show abstract

Section: Consensus and Binding Stepmentioning

confidence: 99%

Section: Consensus and Binding Stepmentioning

confidence: 99%

Section: Consensus and Binding Stepmentioning

confidence: 99%