Which Of The Following Phases Comes Before Orchestrating Services?

Orchestration Layer

Future Trends

Gary Lee , in Cloud Networking, 2014

Information center automation

The ultimate goal for large data centers is the availability of a central orchestration layer that can be used by the data center administrator to classify resources, monitor operations, and quickly recover from error conditions. This includes the orchestration of all data center resource including servers, virtual machines, networking, security, and storage. It is expected that this type of centralized command volition start to be implemented over the next several years as technologies such as software-defined networking (SDN) and network role virtualization (NFV) get more than mature.

Once this centralized orchestration layer is established, it is expected that other types of innovation volition commencement to take place. 1 area is the utilize of information center automation running on top of the orchestration layer. Considering all data eye monitoring information comes dorsum to the orchestration layer, why not automatically utilize this information to adapt the data center configuration without transmission intervention. For case, assume a congestion point starts to develop within ane section of the data center network. The data eye automation layer could quickly recognize this status, move virtual machines, and reroute traffic to eliminate this congestion in the fraction of the fourth dimension information technology would take a human operator. This would not but reduce data center operating expenses, but too amend information center utilization, which in turn could increase information center revenue. We expect that the development of these types of applications will be the side by side area of software innovation for cloud data centers.

Read full chapter

URL:

https://www.sciencedirect.com/scientific discipline/commodity/pii/B9780128007280000114

IoT protocols, architectures, and applications

Chiara Buratti , ... Haibin Zhang , in Inclusive Radio Communications for 5G and Beyond, 2021

7.5.i Software-divers IoT networks

A research activity has been dedicated to the design and test via experimentation of an SDN-based architecture for IoT networks.

The proposed architecture is reported in Fig. seven.21 [CBC⁺17,DCT⁺18] . It is based on the principles of the ETSI NFV Direction and Orchestration (MANO) framework, where high-level components, such equally the orchestration layer, are in charge of managing the end-to-end network function life cycle and orchestrating the resources required to deploy a given service. The orchestration layer must then interact with domain-specific components, that in the instance of the IoT domain include (see Fig. seven.22): i) Management Plane: IoT Virtualized Infrastructure Manager (VIM), which manages resources in the IoT domain; ii) Control Plane: SDN controller, responsible for proper traffic steering, contributing to the E2E service deployment across the domain; iii) Data Aeroplane: IoT gateway and devices, representing the computing and networking resources to be managed and controlled by the other components.

The architecture also includes a database, containing the descriptors of all the IoT devices, including the IP addresses of the gateways enabling the connectivity of each IoT device, the services that they may provide and the related QoS that can be guaranteed. When a asking comes from a user, the IoT VIM maps the incoming service asking to the most suitable Virtual Network part (VNF); in one case having identified the best match, the controller will: i) program the selected IoT network ensuring the requested QoS requirements; and ii) frontward the request to the gateway associated to the target VNF. The main novelty, west.r.t. previous works, such as [GMMP15] and [MR11], relies on the use of an IoT controller capable of programming the IoT networks responding to the QoS requirements specified past the end user.

In [DCT⁺18], the architecture is characterized in terms of the RTT functioning metric. In particular, it is considered the setup shown in Fig. vii.23, where a user periodically asks for data measured by the three unlike sensor nodes connected to the IoT network. These nodes are placed at ane, two, and three hops from the coordinator, respectively. For each packet we measure: i) the RTT at the data plane, that is the interval of time betwixt the inflow of the query coming from the controller, at the application layer of the coordinator, and the arrival of the respond from the target node, again at the application layer of the coordinator; ii) the RTT at the command plane, that is the time interval betwixt the arrival of the query coming from the VIM, at the controller, and the arrival of the reply coming from the gateway, over again at the controller; iii) the RTT at the management airplane, calculated as the difference between the fourth dimension stamps taken when a query arrives to the IoT VIM and when the response to the aforementioned query (if present) is sent dorsum to the requesting user. Average RTT values are reported in Fig. 7.24. As expected the RTT slightly increases with the number of hops and when moving from information, to control, to management planes (each layer adds some processing). Results show the feasibility of the proposed solution, reporting finish-to-end delays at the user in the order of 100 ms.

Read total affiliate

URL:

https://www.sciencedirect.com/science/article/pii/B9780128205815000134

Software-Defined Networking

Gary Lee , in Cloud Networking, 2014

Software-defined information center

The ultimate resource management scheme for cloud data center administrators is the software-divers data eye. What this means is that a unmarried software interface is used to hands configure the data center at a high level with much less human being intervention. For example, a single administrator bug a prepare of commands to allocate virtual machines, virtual networking, and virtual storage to a new tenant with specified service level guarantees, and information technology is completed within a few minutes. Today, three separate administrators may demand to be notified of the new tenant requirements. Each will access multiple resource allotment tools from unlike vendors and attempt to coordinate their efforts. This can be an error-prone procedure that tin can have many days to complete. Effigy 9.1 shows a simplified view of how a software-defined data center tin can amend this situation.

At the top level is the orchestration layer using software such as the OpenStack cloud operating organization. The orchestration layer northbound API can support multiple applications from third parties. For example, an awarding could monitor traffic patterns in the network and automatically adjust virtual network connections or move virtual machines in order to optimize overall network performance. Another application could allocate certain virtual machines and configure dedicated network bandwidth to move storage information to a storage archive during off-acme hours. In order for this new software-divers data center to configure these various resources, the orchestration layer southbound API sends the advisable commands to different controller software modules.

Below the orchestration layer, there are dedicated controller software modules for servers, storage, and networking. These controllers translate commands from the orchestration layer into industry standard open APIs such as OpenStack for server virtualization and OpenFlow for networking. Controllers using open northbound and southbound APIs allow for ease of integration with orchestration layers and also allow use of standard high volume hardware from a variety of vendors that back up the open API. In this example, the hypervisors and virtual switches in the servers must understand commands from the open VM controller and the height of rack switch operating system must understand commands from the open network controller. For storage, an open storage controller can be used to transport commands to configure and monitor the storage resource. By having these controllers communicating with the same orchestration layer, coordination among resources can be guaranteed. In addition, these controllers aid expose bachelor hardware resources and features to the orchestration layer. In the side by side sections, we will provide more than details backside the OpenStack and OpenFlow standards.

Read full affiliate

URL:

https://www.sciencedirect.com/science/article/pii/B9780128007280000096

Network Function Virtualization

Ken Grey , Thomas D. Nadeau , in Network Part Virtualization, 2016

NFV at ETSI

The attempt by network operators (largely from the telco community) to define NFV architecture is covered in detail in Chapter 3, ETSI NFV ISG. Be forewarned—it is an ongoing story. While many of its contributors may refer to ETSI NFV as a "standard," it is (at best) an "incomplete" compages.

In our original description of the action we pointed with hope to 3 workgroups as the keys to success of the compages effort—MANO, INF, and SWA (Fig. 1.8).

Today, fifty-fifty though desired functionality has been divers, much of the necessary practical definitions expected in these architectural components remains undone—particularly in the challenge areas of performance, resource direction, high-availability, and integration.

Some of these may be settled in a time to come phase of ETSI work, but a timing result for early on adopters looms.

Direction and Orchestration (MANO) work did not lead straight to new management tools to broaden traditional OAM&P models to handle virtualized network services. Affiliate half-dozen covers the oversights in NFV Orchestration related to interfaces with legacy OSS/BSS and the complete lack of Service Assurance. Private vendors and their customers are sorting through this today. The roles of diverse managers (VIM and NFVO) are even so existence debated/divers in an ongoing phase of ETSI work.

Resource direction that was envisioned as part of the orchestration layer was defined in terms of capacity just not capability—the latter a problem that emerges in the heterogeneous environments described in Chapter 7, The Virtualization Layer—Performance, Packaging, and NFV and Chapter 8, NFV Infrastructure—Hardware Evolution and Testing.

Minimization of context switching happened organically every bit a requirement to expand Intel's marketplace (previously described), through research projects in academia uncoupled from the influence of ETSI NFV Infrastructure (INF) workgroup and through competition. The identification of a "all-time practice" has proven to be a moving target as a upshot.

ETSI Software Compages (NFV SWA) published piece of work remained VM-centric. While they chanced into describing the decomposition of functions into subcomponents they did not ascertain the anticipated descriptors that would used to communicate the relationships and advice between the parts, operational behavior, and constraints (eg, topology). The approach hither was anticipated to be model driven, and follow-on "open source" activities in MANO take attempted to define these. A number of examples such as the Open-O effort. More details on these are given in Affiliate 7, The Virtualization Layer—Functioning, Packaging, and NFV.

These anticipated part descriptors would act like the familiar concept of policies attached to SDN "applications" as expressed-to/imposed-on SDN controllers. Such policies specify each application's routing, security, operation, QoS, "geo-fencing," admission control, consistency, availability/disaster-recovery, and other operational expectations/parameters.

Such descriptors were to be defined in a style that allows menstruation-through provisioning. That is, while the orchestration system and SDN controller can collaborate on placement, path placement and instantiation of the network functions, their individual and blended configurations can be quite complex—and are currently vendor-specific.

Much of this ultimately fell through to work in OpenStack, YANG modeling in other SDOs, projects in OPNFV or is still undefined.

Some basic attempts to inject policy in the NFV process came well-nigh through the Group Based Policy projection in the OpenDaylight Projection and an experimental plugin to OpenStack Neutron.

Read full chapter

URL:

https://world wide web.sciencedirect.com/scientific discipline/commodity/pii/B9780128021194000018

End-to-finish programmable, cloud-based virtualized HetNet: Advances made & challenges to address

Thou. Moshiur Rahman , ... Sofiène Affes , in Computer Communications, 2017

6.8 Wireless spectrum virtualization

Radio spectrum is the canteen neck for wireless networks, specially those operating in licensed spectrum ring. Hence, licensed spectrum should be virtualized, so that, different VNOs can synchronously share them in fourth dimension (TDMA), frequency (FDMA) and space (SDMA). Spectrum manager in the management and orchestration layer is responsible for managing the sharing of the licensed spectrum among the incumbent VNOs. To mitigate the spectrum scarcity trouble, unlicensed spectrum should too be used in opportunistic style wherever possible. Leveraging cognitive radio (CR) [124] technologies, VNOs tin opportunistically share the gratuitous frequency band that is not currently being utilized past the master users. Discussion of CR technology is out of the telescopic of this paper, interested readers can read the above mentioned newspaper and the references within. Opportunistic spectrum use can exist administered past the Local controllers in the transmission nodes which have admission to spectrum availability information in the surface area where the node is operating.

A spectrum virtualization layer (SVL) is introduced in Ref. [125] that supports flexible spectrum programmability beneath traditional wireless concrete layer. Yang et al. [126] proposes an opportunistic spectrum sharing based resource allotment scheme for virtual wireless networks. The authors codify the wireless resource resource allotment trouble every bit NP-hard integer problem and advise dynamic programming and heuristic algorithms to solve the problem. In Ref. [12] wireless spectrum virtualization was achieved past scheduling spectrum resource to various virtual operators according to spectrum availability and demand from the operators.

Read full article

URL:

https://www.sciencedirect.com/science/article/pii/S0140366417300907

A survey of application orchestration and OSS in next-generation network management

Guy Saadon , ... Noemie Simoni , in Computer Standards & Interfaces, 2019

2.3 Metro Ethernet Forum

Metro Ethernet Forum (MEF), mostly supported by vendors similar ALU, Huawei, CISCO, and AT&T, was founded in 2001 for support of layers two and 3 standard solutions, with specifications and certification processes. MEF at present closes a gap at a higher layer of SDN architecture. The Third network and Lifecycle Service Orchestration (LSO) [27,28] sees the network equally a NaaS for the end user and enables him to create, modify, and delete, dynamically and on-demand, services via Client Web portals or SW applications. The forums aim is to replace the rigid OSS and BSS earth and its silos arroyo [29] with horizontal Orchestration layers in the newer SDN approach. As mentioned in [28], LSO is an agile arroyo to streamlining and automating the service lifecycle in a sustainable fashion for coordinated direction and command across all network domains responsible for delivering an end-to-cease Connectivity Service. Reference compages (see Fig. 5) describes the functional management entities needed to support LSO and the interfaces between them. LSO is related to Orchestration and provides open and interoperable automation of direction operations over the unabridged lifecycle of Layer 2 and 3 Connectivity Services. This includes design, fulfillment, command, testing, problem management, quality management, billing & usage, security, analytics, and policy capabilities over the network domains, which require coordinated management and control in order to deliver the service. LSO MEF reference functional architecture is clearly defined with its limits and interfaces with other layers, (every bit shown in the figure below) and other partner domains. The LSO explains the different roles and functions of the Customer Awarding Coordinator (CUS), of the Concern applications (BUS), and of the Service Orchestration Functionality (SOF) within the same Services Provider (SP) domain and other partner domains and their respective interfaces, such equally CANTATA, SONATA, ALLEGRO, LEGATO, and INTERLUDE.

MEF develops an open and standard architecture that should be able to support layer 2 and layer iii services on demand with complete automation. The MEF Unite plan since 2014 coordinates internal and external engagement with Standards Development Organizations, Associations, and Open Source Projects, to lead the industry migration to orchestrated services. Even though LSO MEF specifies every layer of the SDN architecture, information technology as well gives preference to other forum partners like ONF or TMF for complete network support. Its objective is to define a detailed services Orchestration via functional assay and architecture. Simply LSO MEF still stays very general and the way Orchestration will synchronize and manage the dissimilar applications is not clear.

Read full article

URL:

https://www.sciencedirect.com/science/article/pii/S0920548917304014

A comprehensive survey of Network Role Virtualization

Bo Yi , ... Min Huang , in Reckoner Networks, 2018

1 Introduction

Currently, virtually traditional networks are full of diverse proprietary hardware appliances which are as well called middle-boxes [1] such as firewall and Network Address Translator (NAT). A given service usually has a strong connection with some specific middle-boxes. For example, launching a new service needs to deploy a variety of eye-boxes and accommodating these middle-boxes is becoming more than and more difficult. In addition, designing proprietary hardware based protocols and deploying proprietary hardware are extremely hard, expensive and time-consuming. One typical case is that the procedure of transforming IPv4 to IPv6 has continued for over ten years, and notwithstanding IPv4 is however used widely. Thus, it is extremely difficult to update a protocol running on proprietary hardware, and let alone deploying a new one [two] and [3]. Moreover, with the ever increasing and various service requirements, service providers have to calibration upwards their physical infrastructure periodically, which directly leads to high Majuscule EXpenditure (CAPEX) and OPeration EXpenses (OPEX) [iv].

The Commercial-Off-The-Shelf (COTS) network equipment (e.thousand.,  × 86 based hardware), which can satisfy the needs of full general employ rather than customized purposes, are providing far more capacities with less cost than specialized network equipment. Hence, the COTS hardware has come up as a highly competitive force against defended hardware. In this way, most Telecommunications Operators (TOs) look forward to separating network functions from the purpose-congenital devices and implementing them as software which could be deployed on standard COTS hardware. Under this situation, over twenty of the world's largest TOs, such equally American Telephone and Telegraph (AT&T), British Telecom (BT) and Deutsche Telekom (DT), formed an Industry Specification Group (ISG) within the European Telecommunications Standards Institute (ETSI) to define Network Office Virtualization (NFV) in October 2012 [v] (the acronyms ETSI and ETSI NFV ISG are used synonymously hereafter and all the acronyms used in this work are summarized in Table one). Due to the separation of network function from hardware, NFV can effectively reduce the CAPEX and OPEX. Since then, the ETSI has grown to a large community with 300+ members all over the world including 38 of the world'south major service providers. These members work intensely to develop the required standards for NFV likewise every bit share their experience of NFV implementations and testing [half dozen] and [seven].

Table 1. Acronyms used throughout this work.

Abbreviations	Full proper name	Abbreviations	Full name	Abbreviations	Full name
3GPP	tertiary Generation Partnership Project Service	ATIS	Alliance for Telecommunications Industry Solutions	AT&T	American Telephone and Telegraph
BBF	Broadband Forum	BNG	Broadband Network Gateway	BSS	Business organisation Support Organization
BT	British Telecom	CAPEX	Capital Expenditure	CG-NAT	Carrier Grade Network Address Translator
COTS	Commercial-Off-The-Shelf	CDN	Content Distribution Network	CPE	Client Premise Equipment
C-RAN	Cloud RAN	CS/MG-CF	Call Session/Media Gateway Control Office	DAS	Direct Fastened Storage
DDoS	Distributed Denial of Service	DHCP	Dynamic Host Configuration Protocol	DMTF	Distributed Direction Task Strength
DNS	Domain Name System	DPDK	Data Plane Development Kit	DPI	Deep Packet Inspection
DROP	Distributed Router Open up Platform	DT	Deutsche Telekom	DVR	Distributed Virtual Router
EM	Element Director	EMS	Element Management System	eNodeB	Evolved Node B
EPC	Evolved Parcel Cadre	ETSI	European Telecommunications Standards Institute	FD.io	Fast Information I/O
FM	Menstruation Monitor	FMC	Fixed Mobile Convergence	GA	Genetic Algorithm
GAL	Green Brainchild Layer	GENI	Global Surround for Networking Innovation	GPRS	General Packet Radio Service
GPU	Graphics Processing Unit of measurement	G/South-GSN	Gateway/Serving GPRS Supported Node	HDD	Hard Disk Driver
HLR	Domicile Location Register	HSS	Abode Subscriber Server	IaaS	Infrastructure as a Service
ICN	Information-Axial Networking	IDS	Intrusion Detection System	IETF	Internet Engineering Chore Force
ILP	Integer Linear Programming	IMS	IP Multimedia Subsystem	IoT	Cyberspace of Things
IPS	Intrusion Prevention Organization	IRTF	Internet Inquiry Task Forcefulness	ISG	Industry Specification Group
KVM	Kernel-based Virtual Motorcar	LTE	Long Term Evolution	MANO	Management and Orchestration
MILP	Mixed ILP	MME	Mobility Management Entity	NAS	Network Attached Storage
NETCONF	Network Configuration Protocol	NFV	Network Function Virtualization	NFVI	NFV Infrastructure
NFVIaaS	NFVI as a Service	NFVO	NFV Orchestrator	NFVRG	NFV Inquiry Grouping
NIC	Network Interface Card	N-PoP	NFVI Indicate of Presence	NSH	Network Service Header
NV	Network Virtualization	OEO	Optical-Electricity-Optical	ONF	Open Networking Foundation
OPEN-O	Open Orchestrator	OPEX	Operations Expenses	OSM	Open Source MANO
OSS	Operational Support Organisation	OTN	Optical Transport Network	OVF	Open up Virtualization Format
OVS	Open vSwitch	PaaS	Platform as a Service	PNF	Concrete Network Function
RAN	Radio Access Network	RHEL	Cerise Hat Enterprise Linux	RNC	Radio Network Controller
SA	Simulated Annealing	SaaS	Software as a Service	SAL	Service Availability Levels
SAN	Storage Expanse Network	SDC	Software Defined Compute	SDN	Software-Defined Networking
SDO	Standard Evolution Organizations	SDS	Software Defined Storage	SEG	Security Proficient Group
SFC WG	Service Office Chaining Working Group	SFP	Service Function Path	S/K-GSN	Serving/Gateway GPRS Support Node
SLA	Service Level Agreement	Southward/P-GW	Serving/Public information network GateWay	SR-IOV	Unmarried Root I/O Virtualization
SSD	Solid Land Deejay	TaaS	Tap as a Service	TE	Traffic Engineering
TO	Telecommunications Operators	vE-CPE	Virtuallized Enterprise CPE	VIM	Virtualized Infrastructure Manager
VLAN	Virtual Local Area Network	VM	Virtual Machine	VMM	Virtual Auto Monitor
VNE	Virtual Network Embedding	VNF	Virtual Network Part	VNFaaS	VNF every bit a Service
VNFC	VNF Component	VNF-C	VNF Chaining	VNF FG	VNF Forwarding Graph
VNFM	VNF managing director	VNF-1000	VNF Migration	VNF-MC	VNF Multicast
VNF-P	VNF Placement	VNF-South	VNF Scheduling	VNPaaS	Virtual Network Platform as a Service
VPE	Virtualization Polling Engine	VPN	Virtual Private Network	ZOOM	Aught-fourth dimension Orchestration, Operations and Management

NFV transforms the style that TOs build network past utilizing standard IT virtualization technologies, that is, consolidating various types of proprietary network equipment onto COTS based high volume equipment [8]. Based on the current evolution of virtualization technologies, the advent of NFV makes it possible for most TOs to achieve strong network flexibility and fast new service deployment bike. In this way, TOs can satisfy the continuously growing customer requirements easily and reduce the network functioning and maintenance cost at the aforementioned time. Yet, challenges are ever accompanying opportunities. The network flexibility is achieved by introducing the virtualization plane, which may result in many new problems such as security and scalability.

In principle, all network functions and other network elements tin can be considered for virtualization. These virtualized instances are referred to every bit Virtual Network Functions (VNFs) in the context of NFV, which provide the same functionalities as the corresponding physical instances. Besides, VNFs can be instantiated, executed and deployed by service providers in the NFV Infrastructure (NFVI) environment which provides the required resources (e.g., compute and storage). Typically, chaining multiple VNFs in a particular club can constitute a specific service. To provide the VNF constituted services, about enterprises play more than similar service consumers, because they tin utilize resources in a pay-per-use manner instead of purchasing, configuring, and deploying the infrastructure. In add-on, NFV enables service providers with sure extent flexibility, such that they tin can accommodate the resource allocated to VNFs to satisfy the dynamically irresolute workload of VNFs. This machinery promotes network resources utilization and the agility of network service provision [9].

Currently, at that place are many literatures researching NFV and we categorize them into iii kinds according to their focus. The get-go kind of literatures focused on the integration of NFV and other paradigms. For example, Munoz et al. [10] and Nejabati et al. [11] practical NFV into the optical networks, Omnes et al. [12] practical NFV into the Internet of Things (IoT) and Akyildiz et al. [13] Yang et al. [xiv] and Hawilo et al. [xv] practical NFV into the 5G supported networks. Although these literatures explored many potential of NFV, they directly adopted the standard NFV structure without any modification. All the same, the standard NFV construction was proposed every bit a generic concept. In other words, the standard NFV structure was not specially designed for i specific network scenario and the direct usage of it might atomic number 82 to many unexpected issues which were non discussed in the above works. Too, many works also targeted on integrating NFV with Software Defined Networking (SDN) [xvi–19], cloud computing [20] and [21], etc. Due to the loftier complementary features betwixt SDN and NFV, their integration has been widely recognized. However, not all the integrations with NFV are easy. For case, the integration betwixt NFV supported systems and traditional systems may suffer from many compatibility issues. Despite the fact that some technologies in cloud can be used to accelerate the NFV evolvement, the design of NFVI should be prevented from falling into the design of Infrastructure every bit a Service (IaaS) model of cloud. The second kind of literatures focused on algorithms of many hot topics in NFV, for example, the VNF placement, scheduling and migration. All the existing algorithms tin be classified as either exact ones or heuristic ones. It is known that the verbal algorithms offer optimal solutions which are largely limited by the network scale [22]. Although the solutions offered by heuristic algorithms are not optimal, they are not express by the network calibration [23]. Due to the inherent features of exact algorithms, more and more heuristic algorithms are designed to offer near-optimal solutions with small execution fourth dimension.

The last kind of literatures focused on NFV surveys and reviews (e.1000., Refs. [24–28]). However, some of them simply introduced specific aspects of NFV. For example, Han et al. [25], Mijumbi et al. [26] and Contreras et al. [28] surveyed NFV challenges in terms of innovation, management and functioning respectively, while they failed to present a comprehensive review of NFV to the persons who were unfamiliar to it. In improver, some works intended to present a survey of NFV in other scenarios, for example, NFV in 5G [27]. Unfortunately, Abdelwahab et al. [27] focused more on 5G instead of NFV. It is true that some works indeed presented a relatively comprehensive survey on NFV (e.g., [24]). Mijumbi et al. [24] not merely explained the basic knowledge of NFV, merely also compared NFV with other pop concepts to highlight the business model of NFV. However, since the attention is shifted from hardware to software, the key algorithmic aspects of NFV hot topics are not discussed in Ref. [24]. Therefore, a diverse and comprehensive survey of NFV is still desired.

In this piece of work, nosotros not only present a complete and detailed overview of NFV, but also summarize the pop VNF related algorithms in terms of VNF placement, scheduling, migration, chaining and multicast. Besides, the challenges for NFV are discussed in a bottom upwardly way and the future research directions and awarding scenarios of NFV are also discussed. The main contributions of this work are summarized as follows:

•: Considering the important role that NFV may play in the future, we summarize the existing works with respect to the motivation, terminologies, standardization activities, history, architecture, NFV use cases and solutions in society to provide a comprehensive and detailed introduction for researchers who are new to NFV. In particular, the NFV architecture is presented in a lesser up mode, which includes the physical infrastructure, virtualization layer, virtual infrastructure, management and orchestration layer, and VNF layer.
•: Decoupling network functions from proprietary hardware and implementing them as VNFs outcome in a situation that more and more efforts are contributed to the design and implementation of VNF related algorithms. Apparently, VNF algorithms play an of import role in the evolvement of NFV. Nevertheless, to the best of our knowledge, there is no tutorial and review work near VNF algorithms. Therefore, in this paper, we present an extensive and in-depth discussion of the VNF related algorithms in terms of many popular aspects which include VNF placement, scheduling, migration, chaining and multicast.
•: Although a lot of experiences accumulated during the NFV evolvement and deployment, we should be enlightened that at that place are still many obstacles required to be overcome before using NFV in product networks. Thus, to aid address these obstacles and avert pitfalls as far every bit possible, we first discuss the major challenges that NFV might face, and introduce the related experience that might be used to address such challenges. In item, these challenges are illustrated from lesser up, which include hardware pattern, VNF deployment, VNF life wheel command, service chaining, performance evaluation, policy enforcement, free energy efficiency, reliability and security.
•: NFV can implement its goals independently. Even so, the combination with other popular concepts (e.grand., SDN) would bring significant benefits. Currently, the integration of NFV and SDN is leading a trend towards network softwarization. Thus, we first discuss the futurity directions of NFV in terms of software defined infrastructure, control and application respectively. Then, with the appearance of new paradigms such as 5G and Cyberspace of Things (IoT), network softwarization is alluring more than and more attending. In this style, nosotros likewise discuss the opportunities and challenges that network softwarization may bring to these new paradigms.

The outline of this paper is shown in Fig. i and the rest of this newspaper is organized as follows. We beginning by introducing some related background cognition to help build a preliminary view of NFV in Sections 2 and 3, which include the motivation for NFV, basic terminologies, standardization activities and an evolving history of NFV. Section 4 presents an extensive and comprehensive overview of the NFV reference architecture defined past ETSI, and introduces several typical NFV examples besides as the solutions. Section 5 summarizes the algorithms for many pop issues related to VNF. The ongoing research challenges and future directions of NFV are presented in Sections 6 and dummyTXdummy- 7 respectively. Finally, we summarize this work in Department viii.

Read full article

URL:

https://www.sciencedirect.com/science/article/pii/S1389128618300306

Systematic methods for organising patterns for the internet of things: A preliminary exploration

Vusi Sithole , Linda Marshall , in Internet of Things, 2020

4.3 Organised by attributes and semantics

Another organisation scheme used to group the IoT patterns is co-ordinate to their typical attributes. This method is primarily based on the characterisation of a pattern given its functions and intent, and usually result in several clusters containing patterns that share the same attributes. This system can aid in grouping together a family of related patterns that share some attributes [64]. This method is as well an assist for traversing and learning the nature of the IoT design patterns, as it helps u.s.a. to grasp and understand the role of each blueprint pattern from a content standpoint. Using this attributes-oriented approach, patterns are organised every bit content-driven objects rather than 'difficult-coded, structured elements within layers'. In other words, the focus is on content semantics [65] ¹² that give a rich contexualisation of what the pattern is intended for. That is, each pattern is described using a content model which then becomes an orchestration layer for structural schemas and metadata [66]. In do, the task of characterising each IoT pattern is very demanding and time-consuming. This is mainly due to the fact that the attributes of a pattern are generally captured in several dissimilar documents, with each author using their own choice of terminology to describe the essence of the pattern [44]. Equally such, there is a need to standardise the expression of each pattern'due south attributes to institute commonality [67]. This allows usa to have structured content for each pattern which interpolates the significant of the pattern. In this mode, we are combining structured content with semantics to model each pattern.

In the literature, diverse semantic models are used to organise the concepts that are representative of the IoT patterns attributes [68–lxx]. Some of the semantic models that are used for this task include a taxonomy [71], a thesaurus [44] and an ontology [67]. The subsequent sub-sections below discuss how these semantic models are used to organise the IoT patterns by virtue of their attributes:

4.iii.one Taxonomy

Taxonomies are vocabularies used in several fields to organise content. Structurally, a taxonomy is a hierarchical framework, or schema, for representing events, concepts, and inanimate objects [72]. Graphically represented, they unremarkably resemble a tree turned upside down so that the root is the topmost element. In the IoT, a taxonomy, or more specifically, a concept lattice has been used as a content resource for representation of keywords that are characterisitic of the IoT patterns [71]. These taxonomies are obtained past splitting and reducing general normally circuitous documents, into definitive concepts, which are progressively more than and more than specific [73,74]. In this case, a taxonomy is used as a functional decomposition to model distinct concepts that draw the core essence of the pattern. In this blazon of taxonomy, keyword concepts are classified into three distinct classes, namely: (i) the predicate, which represents the word or descriptor that makes a pattern unique from a family of related patterns. Information technology represents a unique office of that pattern; (ii) intermediate concepts, which are lower-level descriptors of a blueprint that characterise the pattern at a more than coarse-grained level; and the (iii) generic concept, which represents the generic meaning of the pattern, and is a classifier for a family of related patterns [71]. These taxonomies permit for a greater caste of precision in the classification procedure, in which patterns that share the aforementioned generic concepts, and sometimes intermediate concepts, are classified together.

As an instance, Fig. six below presents a hierarchy of concepts that represent the backdrop of a well-known IoT pattern – DIGITAL SIGNATURE. The main idea behind this schema is to characterise each pattern by a main verbal predicate, some intermediate concepts and one generic concept. In Fig. half-dozen, the verbal predicate is identity verification, intermediate backdrop include authentication, cryptography and encryption, and the generic concept for this blueprint is security.

four.3.2 Thesaurus

In view of organising patterns, a thesaurus is used equally a terminological control model for forming a constrained organisation linguistic communication for the IoT cognition domain. Specifically, a thesaurus is used to add dynamic vocabulary of semantically and generically related terms which covers the essence of the given patterns [71]. Since authors generally use complimentary, not-constrained vocabulary when they draw patterns in the IoT domain, it becomes necessary to generate a compilation of words and phrases showing synonyms of all the keywords that characterise each blueprint. The exercise of control on the vocabulary of terms used in indexing the IoT patterns optimises the man value of the content and also enable some type of useful machine interpretation. The IoT, in general, is called by several synonymous names such as the Internet of Everything, Industrial Internet of Things, Web of Things, Industrial Cyberspace, etc. As well, the IoT patterns demand to be contrasted from a conceptual overview in social club to notice their relationships and highlight their differences. In the IoT literature, this is done using several mechanisms such as using the definitions issued by reference organisations and researchers, and making apply of the thesaurus proposed by IEEE, ACM and UNESCO [75].

4.3.3 Ontology

The representational form of collective patterns too as their system in relation to each other constitute the structure of an ontology [44]. Cantankerous-references make explicit the manner in which patterns relate to each other in a network of concepts. Each entry of a pattern in an ontology consists of an entity (which represents an objects or thing, i.e. the pattern name), a relation (which represents relationships between things, refer to Department 4.2), a part (describes the participation of entities in a relation, due east.m. a source edge), and a pack of concepts (which represent the attributes of the pattern). The ontology framework serves to permit united states of america to discover whether the data in the construction has whatsoever inconsistencies (also known as validation) and to logically extract implicit data from information (known every bit inference) [76–78]. In practise, an ontology can exist loaded into a graph to enrich the illustration of relationships between patterns. The office and awarding of an ontology for the organisation of the IoT patterns is detailed in Sithole and Marshall [44] and Wang et al. [79]. In the IoT, several ontologies are used to organise the IoT patterns, outline semantic interoperability, and for general knowledge representation inside the IoT domain. Examples of these ontologies include: IoT-ontology, IoT-O, oneM2M, saref, OWL-IoT-S, spitfire, IoT-lite, ssn, and SA [fourscore]. From these studies, it is evident that ontologies are a powerful tool for organising the IoT patterns.

Read full commodity

URL:

https://www.sciencedirect.com/science/article/pii/S2542660520301025

Software defined networking: Land of the art and research challenges

Manar Jammal , ... Yiming Li , in Computer Networks, 2014

6 Research initiatives for SDN

SDN enables network owners and operators to build a simpler, customizable, programmable, and manageable network. According to the network enquiry community, SDN will change the future of networking and volition import new innovations to the market [86]. With this in mind, a number of enquiry initiatives have proposed SDN prototypes and applied them to DCN, wireless networking, software-defined radio, enterprises, and campus networks.

6.i SDN prototypes

The concept of SDN emerged in 2005, when the authors of [87] proposed a 4D approach to network control and direction. Afterward, a new network compages, Ethane, which provides network control using centralized policies, was described in [88].

Ethane uses a centralized controller that holds network policies to command period routing. It also uses Ethane switches which receive instructions from the controller to forward packets to their destinations. Policies are programmed using a period-based security language based on DATALOG. Ethane was deployed in the Stanford estimator science section to serve 300 hosts and in a small business to serve thirty hosts. Its deployment was an experiment to evaluate central network management, and it showed that a unmarried controller could support 10,000 new flow requests per second for small network designs and that a distributed set of controllers could be deployed for large network topologies. Ethane has two limitations that prevent it from existence implemented using electric current traditional network techniques. Initially, it requires noesis almost network users and nodes, and it demands command over routing at the flow level [89]. These limitations were addressed by NOX, a network operating-system framework.

Under NOX, applications can access the source and destination of each event, and routing modules can perform constrained routing computations. NOX makes it possible to build a scalable network with flexible control considering it uses flows as its intermediate granularity [89].

6.ii Cloud computing and virtualization in SDN

Other contempo studies [90] have adult an SDN-based controller framework, Meridian, for deject-computing networks. Peak provides a network services model that enables users to construct and manage a suitable logical topology for their cloud workloads. In addition, it allows virtual implementations on the underlying physical networks. Inspired by SDN, Elevation is composed of three logical layers: the network model and API layer, network orchestration, and interfaces to network devices. The first layer provides interaction with the network through declarative and query APIs; the declarative API creates the shape of the multi-virtual machine awarding, while the query API supports requests for topology views and network statistics. The orchestration layer provides services such as a global view of the information-center topology, routing algorithms, and scheduling network configuration and control functions. The everyman layer is responsible for creating virtual networks. In addition to the importance of Summit in supporting a service-level model, information technology is considered as an initial image of SDN in the deject. Researchers would like to explore the performance of Meridian in cases of sensitive workloads, the scalability of this framework to support large networks, and its ability to recover failed plans [xc].

six.iii SDN tools and languages

Various tools and languages are used to monitor and implement SDN. Certain SDN initiatives have focussed on a forming platform, Onix, to implement SDN controllers as a distributed system for flexible network management [91]. Other studies have presented a network debugging tool, Veriflow [92], which is capable of discovering the faults in SDN application rules and hence preventing them from disrupting network performance. Additional initiatives [93] take developed a routing architecture, Routeflow, which is inspired past SDN concepts and provides interaction between commercial hardware performance and flexible open up-source routing stacks. Hence, information technology opens the door to migration from traditional IP deployments to SDN.

In addition to recent studies that developed physical SDN prototypes, other researchers [64] accept provided an efficient SDN innovation, Mininet. Mininet is a virtual emulator which provides an environment for prototyping any SDN thought. Whenever the epitome evaluation is acceptable, then information technology tin be deployed in enquiry networks and for full general use [64]. Yet, Mininet'southward services are hindered by certain limitations: poor performance at high loads and its lightweight virtualization arroyo.

Enquiry has as well been directed toward developing control support for SDN and describing new language approaches to program OpenFlow networks.

Foster et al. [74] proposes a design for Frenetic, a high-level linguistic communication for programming OpenFlow architectures. Frenetic consists of a query language based on SQL syntax, a stream-processing language, and a specification language for bundle forwarding. With the combination of these three languages, Corybantic simplifies the programmer'south task by enabling him/her to produce forwarding policies as high-level abstractions.

Information technology addresses some of OpenFlow's shortcomings which are due to the lack of consistency between installing a dominion in the switches and allowing other packets to exist candy, in addition to the lack of synchronization betwixt the packet arrival fourth dimension and the rule installation time. Information technology consists of ii abstraction levels, the source-level operators that deal with network traffic, and the run-time system responsible for installing rules into switches.

In addition to the Frenetic language that can plan OpenFlow networks, a number of other OpenFlow programming languages have been proposed, such as Procera [94,95] and Nettle [60]. These languages are based on functional reactive programming, facilitate network management, and support consequence-driven networks.

six.4 SDN vendors

Ref. [96] describes the Floodlight controller platform. It is an enterprise-grade, Apache-licensed, Java-based OpenFlow controller that supports OpenStack orchestration and virtual and physical switches and manages OpenFlow and non-OpenFlow networks. In add-on, NEC has designed a network virtualization compages encapsulated as NEC ProgrammableFlow. The ProgrammableFlow applied science provides management of their networking cloth. NEC has created custom concrete switches, PF5240 and PF5820, to facilitate the ProgrammableFlow network architecture. The ProgrammableFlow controller can command whatever ProgrammableFlow or OpenFlow switch in a virtual network [97]. Ref. [98] provides an option list of existing OpenFlow controllers (NOX, Beacon [99], Helios, etc.) and switches (software and hardware options such as Open vSwitch and Pronto) to design SDN prototypes.

Besides these initiatives, researchers and enterprises have designed virtualization platforms for SDN [100]. NICIRA has created a consummate SDN solution: the network virtualization platform (NVP). It can be injected over existing network infrastructure or designed into emerging network fabrics. The NVP system works in collaboration with Open vSwitches that are configured in the hypervisor or used as gateways to legacy VLANs. Network virtualization is tasked to the Controller Cluster. The cluster is an array of control structures running on servers separate from the network infrastructure. Control is separated non only from network devices, merely also from the network itself. Each cluster is capable of controlling thousands of Open vSwitch devices. The NVP architecture combines control and switching abstractions to provide a versatile network solution [101].

Finally, It organizations and enterprises are focusing on applying SDN non simply to information-eye networks (LANs), but likewise to wireless local-area networks (WLANs) and wide-area networks (WANs), where OpenFlow will function as an overlay over L2 and L3 virtual private networks (VPN) [102]. HP has announced that an SDN-centralized controller can minimize the price and complication of implementing WAN optimization schemes. A image of SDN, Odin, was described in [103] and was intended to enable network operators to deploy WLAN services equally network applications. Odin consists of a master, agents, and applications. The master runs as an application on the OpenFlow controller, controls the agents, and updates the forwarding tabular array of access points (APs) and switches, and the agents run on the APs and collect information about the clients.

Read full article

URL:

https://www.sciencedirect.com/science/article/pii/S1389128614002588