Network Function Virtualization Research Articles

Taxonomic Exploration of Healthcare IoT: Challenges, Solutions, and Future Frontiers

An Internet of things (IoT) ecosystem is a fast-developing network in which users can connect a heterogeneity of physical and virtual devices, including customized healthcare areas. As medical resources are scarce, populations are aging with chronic diseases and require remote monitoring, medical expenses are rising, and telemedicine is being demanded in developing nations, the IoT is an attractive topic in healthcare. Through the IoT, people can enjoy better health and diminish pressure on sanitary systems. In this study, previously published studies in Healthcare IoT (HIoT) systems are detailed, analyzed, and taxonomically classified. By categorizing the articles according to the types of HIoT systems, we dispense a detailed taxonomical study. In addition, different evaluation methodologies, tools, and metrics are discussed, along with their advantages and disadvantages. The studies indicate that power management, trust, privacy, fog computing, and resource management are among the open issues. The future of the Internet includes tactile networks, social networks, big data analytics, software-defined networking, network function virtualization, the Internet of nano things (IoNT), and blockchain. It would be beneficial to study and research HioT systems further in terms of interoperability, the implementation of real-world test beds, scalability, and mobility.

On the feasibility of using an autonomic control loop and automated planning for tenant-oriented reconfiguration of network slices

In 5G, 6G and beyond networks, tenants need capabilities to reconfigure network slices since they must react quickly to changing network conditions and end-users’ demands to meet Service Level Agreements. Offering reconfiguration operations to tenants is challenging since those functions must be autonomous and governed by high-level network management policies. Existing approaches face the network slice reconfiguration problem from the provider perspective, disregarding the importance of providing slice reconfiguration capabilities to tenants. This paper presents an approach called ATRAP to demonstrate the feasibility of using an autonomic control loop and automated planning for reconfiguring network slices from the tenant perspective. ATRAP introduces a tenant-oriented architecture based on the monitor-analyze-plan-execute-knowledge method and automated planning for computing reconfiguration plans autonomously. An ATRAP’s plan aims to turn a network from a source configuration where one or more intents representing high-level network management policies are unmet into a target configuration in which the network meets the intents again. As ATRAP represents a network slice as a Service Function Chain comprising Virtual Network Functions (VNFs) and virtual links connecting them, a reconfiguration plan involves tasks to migrate VNFs and virtual links into the substrate network in a particular order. Results revealed as ATRAP computes plans with few migrations of VNFs and their adjacent links, leading to low reconfiguration costs related to short service disruptions. Also, results showed as ATRAP’s planning time depends on the size of the substrate network and the tenant’s slices. The results evidence that ATRAP is an attractive and feasible solution for the in-tenant reconfiguration of network slices.

Joint VNF Parallelization and Deployment in Mobile Edge Networks

Mobile edge computing (MEC) has emerged as a promising computing paradigm that provides flexible and responsive local services for mobile user equipment at the network edge. Software instances for user equipment tasks are typically deployed as Virtualized Network Functions (VNFs) at resource-constrained edge nodes. Task data exchanged across the VNFs in serial can incur high task completion latency. It is therefore desirable to deploy certain VNFs in parallel. However, deciding where to deploy VNFs depends on which VNFs are parallel, and conversely, their deployment also affects their parallel execution. In this paper, for the first time, we jointly consider the parallelization and deployment strategies for VNFs at edge nodes. We closely examine the complexity of the joint optimization problem and introduce an Improved Service Function Graph (I-SFG) that reflects the coordination and dependency relations among the VNFs to provide parallel services for each piece of user equipment. We first propose an approach based on integer linear programming to find optimal solutions in small-scale scenarios and then present an effective solution through cascading I-SFG construction and VNF deployment approximation to solve large-scale problems. Theoretical analyses and experimental results show the superiority of our joint design and the proposed practical solution.

DRIVE: Dynamic Resource Introspection and VNF Embedding for 5G Using Machine Learning

Network Slicing (NS) technique is comprehensively reshaping the next-generation communication networks (e.g. 5G, 6G). Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) predominantly control the flow of service functions on NS to incorporate versatile applications as per user demands. In the virtualized-Software Defined Networking vSDN environment, a chain of well-defined virtual network functions (VNFs) are installed on Service Function Chains (SFCs) by multiple Internet Service Providers (ISPs) concurrently. Generation, allocation, re-allocation, release and destroying associative VNFs on SFC is an extremely difficult task while keeping high selection accuracy. Towards solving this fundamental issue, in this work, we have proposed a multi-layered SFC formation for adaptive VNF allocation on dynamic slices. We have formulated an ILP to address the VNF-EAP (VNF-Embedding and Allocation Problem) over real network topology (AT&T Topology). Leveraging machine learning techniques we have shown an intelligent VNF selection mechanism to optimize resource utilization. The performance evaluation shows remarkable efficiency on ML-driven dynamic VNF selections over static allocations on SFCs by halving resource usage. Further, we have also studied a VNF typecasting technique for service backup on outage slices in the field of disaster management activities.

Deep Reinforcement Learning for Resource Demand Prediction and Virtual Function Network Migration in Digital Twin Network

Internet of Things (IoT) enables intelligent services varying with the complex and realtime environment to achieve network benefits, where Network Function Virtualization (NFV) can dynamically provide Virtualized Network Functions (VNFs) for IoT devices. In NFV-enabled IoT architecture, a service function chain (SFC) consists of an ordered set of virtual network functions (VNFs). However, the energy consumption of the VNF migration and SFC reconfiguration is one major issue owing to the dynamic characteristic of IoT network. In this paper, we propose a new paradigm digital twin (DT) to create the virtual twin of physical objects in IoT network, then, we formalize the problem as a mathematical model, which aims to minimize the energy consumption. To this end, we prove this problem is NP-hard and propose an algorithm Bidirectional Gated Recurrent Unit (Bi-GRU) based on federated learning to predict the resource requirement. Further more, according to the prediction result, which utilizing the deep reinforcement learning (DRL) algorithm for decision making of the VNF migration. Simulation results show that our proposed method can effectively reduce the number of VNFs to be migrated and economize the energy consumption of the digital twin IoT network.

Function traffic-aware VNF migration for service restoration in an NFV-enabled IoT system

Maintaining a service continuum is crucial for providing a reliable service in an Internet-of-Things (IoT) system based on Network Function Virtualization (NFV) due to the distribution of Virtual Network Functions (VNF). We address the optimization problem of VNF migration in an NFV-enabled IoT system (NIoT), considering the dynamics of traffic volume processed by an individual IoT function. We develop an optimization model based on integer linear programming and a function traffic-aware heuristic algorithm to solve the problem. The evaluation results show that our proposed heuristic algorithm offers an approximation solution very close to the optimal VNF migration in significantly reduced time. It outperforms a benchmark solution based on Simulated Annealing in both the service restoration cost and computation time. Furthermore, our findings provide valuable insights into an investment strategy for an IoT service provider seeking to enhance the system performance in the presence of a failure.

Towards Attack-Resistant Service Function Chain Migration: A Model-Based Adaptive Proximal Policy Optimization Approach

Network function virtualization (NFV) supports the rapid development of service function chain (SFC), which efficiently connects a sequence of network virtual function instances (VNFIs) placed into physical infrastructures. Current SFC migration mechanisms usually keep static SFC deployment after finishing certain objectives, and deployment methods mostly provide static resource allocation for VNFIs. Therefore, the adversary has enough time to plan for devastating attacks for in-service SFCs. Fortunately, moving target defense (MTD) was proposed as a game-changing solution to dynamically adjust network configurations. However, existing MTD methods mostly depend on attack-defense models, and lack adaptive mutation period. In this paper, we propose an Intelligence-Driven Service Function Chain Migration (ID-SFCM) scheme. Firstly, we model a Markov decision process (MDP) to formulate the dynamic arrival or departure of SFCs. To remove infeasible actions from the action space of MDP, we formalize the SFC deployment as a constrained satisfaction problem. Then, we design a deep reinforcement learning (DRL) algorithm named model-based adaptive proximal policy optimization (MA-PPO) to enable attack-resistant migration decisions and adaptive migration period. Finally, we evaluate the defense performance by multiple attack strategies and two realistic datasets called CICIDS-2017 and LYCOS-IDS2017 respectively. Simulation results highlight the effectiveness of ID-SFCM compared with representative solutions.

Securing NFV/SDN IoT Using VNFs Over a Compute-Intensive Hardware Resource in NFVI

The Network Function Virtualization (NFV) and Software-Defined Networking (SDN) are network paradigms for flexibly deploying future networks while guaranteeing security service requirements. This work designs Virtual Network Functions (VNFs) through a Compute-Intensive (CI) hardware resource in Network Function Virtualization Infrastructure (NFVI). The proposed NFVI is characterized by a multi-channel cryptosystem, which can be virtualized as a plurality of VNFs, that is, crypto engines, and each crypto engine is logically dedicated to an NFV/SDN Internet of Things (IoT) device, which does not have or has limited security capability owing to resource constraints. To enhance the performance of a cryptosystem, the accelerator circuit is often deeply pipelined and unrolled. However, to fulfill the popular feedback operation modes, the throughput of the pipelined and unrolled cryptosystem that implements a block cipher, say Advanced Encryption Standard (AES), can deteriorate even lower than that without these techniques. To solve this problem, we design a pipelined and unrolled multichannel cryptosystem, which can be integrated into NFVI edge servers, with feedback operation mode for the NFV/SDN IoT. As a result, the combinational logics of a block cipher with feedback can be shared by plenty of IoT devices to enhance the hardware efficiency as well. Moreover, in addition to briefly review several AES designs, the fastest AES design is derived by shortening its critical path to only a logic gate of multiplexer or Exculsive OR (XOR) gate.

Research on Reliable Deployment Algorithm for Service Function Chain Based on Deep Reinforcement Learning

This paper investigates the reliable deployment algorithm for Service Function Chains (SFC) based on deep reinforcement learning. SFC, as a chained function composition for complex network services, plays a crucial role in improving network efficiency and stability. To address the issue of existing SFC deployment algorithms that overlook the reliability of network functions and links, this paper proposes a deep reinforcement learning-based algorithm that utilizes a virtual network function and virtual link reliability mapping model for optimization. By learning the mapping between system states and actions, the algorithm can optimize the deployment strategy of SFC, thereby enhancing its reliability and performance. Experimental results demonstrate that the proposed algorithm can significantly improve the reliability of SFC and have practical implications for network service deployment.

The 6G Ecosystem as Support for IoE and Private Networks: Vision, Requirements, and Challenges

The emergence of the sixth generation of cellular systems (6G) signals a transformative era and ecosystem for mobile communications, driven by demands from technologies like the internet of everything (IoE), V2X communications, and factory automation. To support this connectivity, mission-critical applications are emerging with challenging network requirements. The primary goals of 6G include providing sophisticated and high-quality services, extremely reliable and further-enhanced mobile broadband (feMBB), low-latency communication (ERLLC), long-distance and high-mobility communications (LDHMC), ultra-massive machine-type communications (umMTC), extremely low-power communications (ELPC), holographic communications, and quality of experience (QoE), grounded in incorporating massive broad-bandwidth machine-type (mBBMT), mobile broad-bandwidth and low-latency (MBBLL), and massive low-latency machine-type (mLLMT) communications. In attaining its objectives, 6G faces challenges that demand inventive solutions, incorporating AI, softwarization, cloudification, virtualization, and slicing features. Technologies like network function virtualization (NFV), network slicing, and software-defined networking (SDN) play pivotal roles in this integration, which facilitates efficient resource utilization, responsive service provisioning, expanded coverage, enhanced network reliability, increased capacity, densification, heightened availability, safety, security, and reduced energy consumption. It presents innovative network infrastructure concepts, such as resource-as-a-service (RaaS) and infrastructure-as-a-service (IaaS), featuring management and service orchestration mechanisms. This includes nomadic networks, AI-aware networking strategies, and dynamic management of diverse network resources. This paper provides an in-depth survey of the wireless evolution leading to 6G networks, addressing future issues and challenges associated with 6G technology to support V2X environments considering presenting +challenges in architecture, spectrum, air interface, reliability, availability, density, flexibility, mobility, and security.

Availability‐aware virtual network function placement based on multidimensional universal generating functions

AbstractNetwork function virtualization (NFV) implements network functions as software, which enables flexible, resource‐efficient, and rapid provision of network services. In NFV, network services are realized by the service function chain (SFC), which is a chain of virtual network functions (VNFs) linked in the proper order. Both availability and deployment cost are key concerns for network operators providing network services as SFC. This paper proposes a flexible VNF placement model on a per‐VNF instance basis that minimizes deployment costs while satisfying availability requirements that may be placed on SFC. This paper uses a multidimensional universal generating function (MUGF) method, which is a multistate system analysis method, to compute the availability of a multistate SFC system with multiple VNFs coexisting on a server. The MUGF method calculates the performance of the entire SFC by combining the performance of servers as determined by applying a continuous‐time Markov chain. To reduce the time to compute the SFC availability, we introduce operators to be applied to MUGF and develop an availability computing method. In addition, a heuristic algorithm for determining VNF placement targeting the lowest deployment cost possible while meeting availability requirements is presented. Numerical results show that the proposed model obtains VNF placement with lower cost than the conventional model in all examined cases. The proposed model achieves VNF placement at 58.5%–75.0% of the deployment cost of the conventional model for the same SFC availability requirements.

Cloud native Lightweight Slice Orchestration (CLiSO) framework

Cloud-native network functions are becoming promising for 5G and beyond networks. They provide the much needed agility and flexibility that was missing from virtual machines. Though, this paradigm shift of using cloud-native application design principles, containerization, microservices, high resilience, and on-demand scaling has created challenges for legacy orchestration systems. They were designed for handling virtual machine-based network functions. Indeed, network slice orchestration requires interaction with multiple technological domain orchestrators, access, transport, core network, and edge computing. The specifications and existing orchestrators are made on top of the legacy virtual machine based network function orchestration. Hence, this limitation constrains their approach to managing a cloud-native network function. To overcome their challenges, we propose a novel Cloud-native Lightweight Slice Orchestration (CLiSO) framework extending our previously proposed Lightweight edge Slice Orchestration (LeSO) framework. In addition, we present a technology-agnostic and deployment-oriented network slice template. To allow zero-touch management of network slices, our framework provides a concept of Domain Specific Handlers. The framework has been thoroughly evaluated via orchestrating OpenAirInterface container network functions on public and private cloud platforms.

A hybrid analytical concept to QoE index evaluation: Enhancing eMBB service detection in 5G SA networks

The launch of commercial 5G networks has unlocked numerous opportunities for heavy data users and high-speed applications. The expected requirements for enhanced mobile broadband (eMBB) are pushing end-users to adopt 5G optimistically. Though already deployed 5G networks have shown high data rates and very low latency, the service-based experience and application behavior have been challenging to monitor. The legacy quality of experience (QoE) and service (QoS) monitoring and evaluation techniques have shown limitations in 5G standalone networks. The current 5G deployment large amount of user plane traffic generated by end-users makes the legacy-monitoring task very costly for mobile network operators (MNOs). And the complexity of the projected future 5G architecture, including advanced technologies such as network functions virtualization (NFV), software-defined networking (SDN), and network slicing, makes traditional service detection and QoE assessment ineffective. In this paper, we discuss a cost-effective hybrid analytical approach to eMBB service detection, analysis, and perceived user QoE measurement from raw traffic in a live 5G standalone (SA) network. We first use flow-level-based packet inspection and machine learning to detect and classify eMBB services from raw traffic. We then use a statistical approach to compute the user quality index (UQI). The concept is tested on traffic captured on a fixed 5G SA network. And the output enabled the MNO to have a 5G QoE assessment structure and awareness to adjust network traffic policies.

VNF Migration in Digital Twin Network for NFV Environment

Network Function Virtualization (NFV) allows for the dynamic provisioning of Virtual Network Functions (VNFs), adapting services to the complex and dynamic network environment to enhance network performance. However, VNF migration and energy consumption pose significant challenges due to the dynamic nature of the physical network. In order to maximize the acceptance rate of Service Function Chain Requests (SFCR), and reduce VNF migration and energy consumption as much as possible, we summarize several related factors such as the node hosting state, link hosting state, energy consumption, migrated nodes, and whether the mapping is successful. We define the Markov decision process by considering the factors mentioned above. Next, we design the VNF migration algorithm utilizing actor–critic models, graph convolution networks, and LSTM networks. In order to reduce the risk of trial and error during training and prediction in deep reinforcement learning scenarios, we designed a network architecture based on a digital twin (DT). In simulation experiments, compared with the FF algorithm that greedily selects the first available node, our AC_GCN algorithm significantly improves the acceptance rate of SFC requests by 2.9 times more than the FF algorithm in small topology experiments, and 27 times more than the FF algorithm in large topology experiments. Compared with the deep reinforcement learning (DRL) algorithm, which does not consider all the above factors together, for the small topology experiment, our AC_GCN algorithm outperforms the DRL algorithm in terms of request acceptance rate by 13%, underperforms compared to the DRL algorithm in terms of energy consumption by 3.8%, and underperforms compared to the DRL algorithm in terms of the number of migrated nodes for 22%; for the large topology experiment, our AC_GCN algorithm outperforms the DRL algorithm in terms of the request acceptance rate by 7.7%, outperforms the DRL algorithm in terms of energy consumption by 0.4%, and outperforms the DRL algorithm in terms of the number of migrated nodes by 1.6%.

HADES: An NFV solution for energy-efficient placement and resource allocation in heterogeneous infrastructures

Network Function Virtualization (NFV) aims to replace traditional network functions running in proprietary hardware with software instances (i.e., Virtual Network Functions, VNFs) embedded in general-purpose virtualization solutions. Aware that the transition to a fully virtualized network infrastructure will pay a high energy cost, especially in IoT systems composed of a myriad of devices, energy efficiency is one of the key innovative targets of future networks. Edge computing should be considered in IoT environments to save time and energy by processing data near the producer devices. However, applying NFV in the context of IoT/Edge/Cloud environments complicates the placement of VNFs, due to the inherent heterogeneous nature of such environments and the variety of resource demands. This paper proposes an energy-aware placement of service function chains of VNFs and a resource-allocation solution for heterogeneous edge infrastructures that considers the computation and communication delays according to the VNFs’ location in the infrastructure. The solution has been integrated with the ETSI-sponsored project Open Source Management and Orchestration (OSM) as an extension called HADES, which allows the configuration of VNFs and their subsequent resource allocation and deployment at the edge, minimizing energy consumption and ensuring a quality of service. We have applied the deployment of augmented reality services in real and simulated scenarios. The results show up to a 59% reduction in power consumption and QoS compliance in all scenarios considered compared to default OSM placement and four other allocation policies. We prove that our solution has negligible power overhead, and validate the scalability and applicability of HADES.

Framework for Virtualized Network Functions (VNFs) in Cloud of Things Based on Network Traffic Services

The cloud of things (CoT), which combines the Internet of Things (IoT) and cloud computing, may offer Virtualized Network Functions (VNFs) for IoT devices on a dynamic basis based on service-specific requirements. Although the provisioning of VNFs in CoT is described as an online decision-making problem, most widely used techniques primarily focus on defining the environment using simple models in order to discover the optimum solution. This leads to inefficient and coarse-grained provisioning since the Quality of Service (QoS) requirements for different types of CoT services are not considered, and important historical experience on how to provide for the best long-term benefits is disregarded. This paper suggests a methodology for providing VNFs intelligently in order to schedule adaptive CoT resources in line with the detection of traffic from diverse network services. The system makes decisions based on Deep Reinforcement Learning (DRL) based models that take into account the complexity of network configurations and traffic changes. To obtain stable performance in this model, a special surrogate objective function and a policy gradient DRL method known as Policy Optimisation using Kronecker-Factored Trust Region (POKTR) are utilised. The assertion that our strategy improves CoT QoS through real-time VNF provisioning is supported by experimental results. The POKTR algorithm-based DRL-based model maximises throughput while minimising network congestion compared to earlier DRL algorithms.

Discrete spider monkey optimization algorithm for latency‐sensitive VNF deployment and resource allocation

SummaryThe next‐generation networks are expected to support diverse latency‐sensitive services, that is, auto‐driving, telemedicine, and so on. Deploying these services in current hardware‐based networks faces significant challenges. The emergence of network function virtualization (NFV) and software defined network (SDN) makes network deployment flexible and agile. In order to meet the strict latency constraints, joint VNF deployment and resource allocation problems are needed. In this paper, we consider the resource‐delay interpendency and formulate the joint VNF deployment and resource allocation problems as linear programming (LP) problems. Discrete spider monkey optimization (DSMO) algorithm is proposed to solve this problem, which imitates spider monkeys' fission and fusion social behavior during foraging. The two‐phase decision update mechanism greatly increases the diversity of the spider monkey decision population and promotes the algorithm to perform a global search of the solution space. Evaluation results show that the proposed solutions achieve superior performance in terms of latency, CPU utilization and service acceptance rate.

Network-Compute Co-Optimization for Service Chaining in Cloud-Edge-Radio 5G Networks

Flexible resource management can be facilitated by recent drastic advances in software-defined networking (SDN) and network function virtualization (NFV) technologies. Most of recent studies on resource management have mainly focused on the independent optimization of network function and computing function on top of a single-layer SDN architecture. In this paper, we focus on networking and processing resource orchestration to enhance the performance of network/computing functions in a hierarchical cloud-edge-radio 5G network architecture. Specifically, we propose a dynamic resource management algorithm, namely <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">RACER</i> , to process network/computing functions by opportunistically exploiting a tradeoff between the characteristics of the cloud and edge with respect to the amount of available resources and transmission delay. Herein, we decompose an original long-term problem including complex decisions in all network layers (that is, cloud, edge, and radio) into spatio-temporal subproblems without loss of optimality by leveraging Lyapunov optimization. Then, we derive the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">RACER</i> algorithm to find the decisions in each layer and each time slot. This online-fashioned algorithm is pragmatic because it not only exploits information obtained from instantaneous time slots, but also operates in a distributed manner with small message exchanges between the different layers. Moreover, we demonstrate the excellent performance improvement of the proposed dynamic and holistic <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">RACER</i> algorithm compared to static and independent policies in the 5G network environment.

SAFE: Service Availability via Failure Elimination Through VNF Scaling

Virtualized network functions (VNFs) enable software applications to replace traditional middleboxes, which are more flexible and scalable in the network service provision. This paper focuses on ensuring Service Availability via Failure Elimination (SAFE) using VNF scaling, that is, given the resource requirements of VNF instances, finding an optimal and robust instance consolidation strategy, which can recover from one instance failure quickly. To address the above problem, we present a framework based on rounding and dynamic programming. First, we discretize the range of resource requirements into several sub-ranges, and thus the number of instance types becomes a constant. Second, we further reduce the number of instance types by gathering several small instances into a bigger one. Third, we propose an algorithm built on dynamic programming to solve the instance consolidation problem with a limited number of instance types. Finally, we set up a testbed to profile the functional relationship between the resource and the throughput for different types of VNFs, and conduct simulations to validate our theoretical results according to profiling results. The simulation results show that our algorithm outperforms the standby deployment model by <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$27.33\%$</tex-math> </inline-formula> on average in terms of the number of servers required. Furthermore, SAFE has marginal overheads, around <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$7.22\%$</tex-math> </inline-formula> , compared to the instance consolidation strategy without VNF backup consideration.

Virtual Network Function Placement Model Considering Both Availability and Probabilistic Protection for Service Delay

This paper proposes a virtual network function (VNF) placement model considering both availability and probabilistic protection for the service delay to minimize the service deployment cost. Both availability and service delay are key requirements of services; a service provider handles the VNF placement problem with the goal of minimizing the service deployment cost while meeting these and other requirements. The previousworks do not consider the delay of each route which the service can take when considering both availability and delay in the VNF placement problem; only the maximum delay was considered. We introduce probabilistic protection for service delay to minimize the service deployment cost with availability. The proposed model considers that the probability that the service delay, which consists of networking delay between hosts and processing delay in each VNF, exceeds its threshold is constrained within a given value; it also considers that the availability is constrained within a given value. We develop a two-stage heuristic algorithm to solve the VNF placement problem; it decides primary VNF placement by solving mixed-integer second-order cone programming in the first stage and backup VNF placement in the second stage. We observe that the proposed model reduces the service deployment cost compared to a baseline that considers the maximum delay by up to 12%, and that it obtains a feasible solution while the baseline does not in some examined situations.