Network Function Virtualization
Description
This book explores the newly emerged technique of network function virtualization (NFV) through use cases, architecture, and challenges, as well as standardization and open source implementations. It is the first systematic source of information about cloud technologies' usage in the cellular network, covering the interplay of different technologies, the discussion of different design choices, and its impact on our future cellular network.
Network Function Virtualization: Concepts and Applicability in 5G Networks reviews new technologies that enable NFV, such as Software Defined Networks (SDN), network virtualization, and cloud computing. It also provides an in-depth investigation of the most advanced open source initiatives in this area, including OPNFV, Openstack, and Opendaylight. Finally, this book goes beyond literature review and industry survey by describing advanced research topics such as service chaining, VNF orchestrations, and network verification of NFV systems. In addition, this resource:
* Introduces network function virtualization (NFV) from both industrial and academic perspectives
* Describes NFV's usage in mobile core networks, which is the essence of 5G implementation
* Offers readers a deep dive on NFV's enabling techniques such as SDN, virtualization, and cloud computing
Network Function Virtualization: Concepts and Applicability in 5G Networks is an ideal book for researchers and university students who want to keep up with the ever-changing world of network function virtualization.
More details
Other editions
Additional editions
Person
Ying Zhang is a Software Engineer in the Facebook network infrastructure team. Prior to joining Facebook, she was a senior researcher and research manager at Hewlett Packard Labs. She has served on the technical program committee of several conferences and workshops, and is a senior member of the IEEE. She has been granted 30+ US/International patents. She was reported by Swedish media as a "Mobile Network 10 Brightest Researcher" in 2015 and was awarded the N2Women Rising Star in Networking and Communications award in 2017.
Content
List of Figures ix
List of Tables xii
Preface xiii
List of Abbreviation and Acronyms xv
1 Introduction 1
1.1 Cloud-Enabled 5G: SDN and NFV 3
1.1.1 Benefits 6
1.1.2 Challenges 7
1.2 Supporting Technologies 8
1.2.1 Cloud Computing 8
1.2.2 Network Virtualization 9
1.2.3 Network Functions Virtualization 9
1.2.4 Software-Defined Networking 10
1.3 Outline of Chapters 10
2 Virtualization and Cloud Computing 13
2.1 Cloud Computing 13
2.1.1 Architecture 13
2.1.2 Types of Clouds 15
2.1.3 Challenges 16
2.2 Host Virtualization 18
2.2.1 Overview 18
2.2.1.1 Benefits 19
2.2.1.2 Use Cases 19
2.2.2 Virtualization Techniques 20
2.2.2.1 Hardware-Level Virtualization 20
2.2.2.2 Other Virtualization Techniques 21
2.2.3 Containers 21
2.3 Network Virtualization 22
2.3.1 Overlay Networks 24
2.3.2 Virtual Private Network 24
2.3.3 Virtual Sharing Networks 27
2.3.4 Switch-Based SDN Virtualization 29
2.3.5 Host-Based Network Virtualization 31
2.4 Wireless Virtualization 32
2.5 Summary 36
3 Network Function Virtualization 37
3.1 NFV Architecture 38
3.2 NFV Use Cases and Examples 42
3.3 NFV Challenges 45
3.4 NFV Orchestration 46
3.4.1 NFV Performance Characterization 47
3.4.2 NFV Performance Improvements 49
3.5 NF Modeling 50
3.5.1 Source-Code-Based Modeling 52
3.5.1.1 Background 52
3.5.1.2 Modeling Example 53
3.5.1.3 Models 55
3.5.1.4 Model Extraction Overview 56
3.5.2 Modeling Applications 58
3.6 VNF Placement 59
3.7 Summary 63
4 Software-Defined Networks Principles and Applications 65
4.1 SDN Overview 66
4.1.1 Motivations 66
4.1.2 Architecture 67
4.1.2.1 Separation of Control and Data Plane 67
4.1.2.2 Northbound Open APIs 69
4.1.2.3 Southbound Control/Data Plane Protocol 69
4.1.2.4 Applications 70
4.1.3 Use Cases 70
4.2 SDN Controller 71
4.2.1 Controller Deployment Choices 73
4.2.2 Apps on SDN Controller 75
4.3 SDN Data Plane 76
4.4 SDN Management 78
4.4.1 Anomaly Detection 78
4.4.2 Network Measurement 80
4.4.3 Failure Recovery 82
4.4.4 Controller Placement 84
4.4.4.1 A Special Study: Controller to Switch Connectivity 87
4.4.4.2 Node Versus Link Failure 92
4.4.4.3 Downstream Versus Upstream Nodes 93
4.5 SDN Security Attack Prevention 94
4.6 SDN Traffic Engineering 96
4.6.1 TE Architecture and Solution Overview 98
4.6.1.1 Hierarchical Control Structure 98
4.6.1.2 Control Node Design 99
4.6.1.3 Scalability Benefit 100
4.6.2 Design Challenges 100
4.6.3 TE Solution Overview 101
4.7 Summary 102
5 SDN and NFV in 5G 105
5.1 5G Overview 106
5.1.1 Architecture 106
5.2 Service Function Chaining 108
5.2.1 OpenFlow-Based SFC Solution 110
5.2.1.1 Represent Directions with Port Types 114
5.2.1.2 Realize Policies with Multiple Forwarding Tables 115
5.2.1.3 Handle Dynamics with the Microflow Table 116
5.2.1.4 Encode Service Chaining with Metadata 117
5.2.1.5 Summary of Dataplane Functions 118
5.2.2 SFC Monitoring 118
5.2.2.1 Handling Multiple Monitoring Tasks 121
5.2.3 Optical SFC 122
5.2.3.1 Service Placement in Optical NFV 125
5.2.4 Verification of Service Function Chaining 128
5.3 Core Network Functions Virtualization: vEPC 131
5.3.1 Existing Solutions Problems 133
5.3.2 Virtualization and Cloud-Assisted PC 134
5.4 Virtualized Customer Premises Equipment 137
5.4.1 Requirements 138
5.4.2 Design 140
5.5 Summary 142
6 Open Source and Research Activities 143
6.1 Open Source Initiatives 144
6.1.1 OpenStack 144
6.1.2 OpenDayLight 146
6.1.3 OPNFV 148
6.1.4 CORD: Central Office Re-architected as a Data Center 150
6.2 NFV Research Problems 151
6.3 Summary 152
References 153
Index 163
Chapter 2
Virtualization and Cloud Computing
2.1 Cloud Computing
Cloud computing has become a widely used computing model to support cost-effective and efficient data processing using commodity servers. Cloud computing makes effective use of distributed environments for tackling large-scale computation problems on vast data set. There are multiple challenges with cloud computing, such as virtualization, isolation, performance, scalability, privacy, and security. In this section, we first provide an overview of the architecture of cloud computing. Then, we will go deeper into various virtualization technologies and focus on the network virtualization.
2.1.1 Architecture
Public cloud providers use an on-demand, pay-as-you-go model of compute and storage infrastructure as well as platform services. Amazon Web Services (AWS) led the early cloud computing revolution, beginning with their S3 service in 2006. Their services have been adopted by companies large and small, from backups and archival storage in S3, to compute in EC2, Virtual Private Clouds, IAM authorization and authentication, and RDS managed databases, to name a few. For customers, these services are easy to add, easy to consume, and can lead to a sprawling, poorly documented infrastructure.
Cloud computing can be viewed as a layering architecture, as shown in Figure 2.1.
- The hardware layer includes the physical resources in the cloud, that is, the hosting facilities, servers, switches, routers, hardware middleboxes, and power and cooling support. The Hardware-as-a-Service (HaaS) model means buying IT hardware or portions of data centers as a pay-as-you-go subscription service. Similar to other cloud computing layers, it shares the benefit of dynamically scaling up and down as the demand changes. The hardware is typically in the form of data centers, which consist of thousands of servers in racks. The HaaS provider needs to handle various hardware management issues such as configurations, fault tolerance, backup powers, and regular maintenance.
- The infrastructure layer is also known as the virtualization layer. The Infrastructure-as-a-Service (IaaS) provides computing resources as a service. Virtualization is an elegant and transparent way to enable time sharing and resource sharing on the common hardware. It allows customers to pay as they grow. By decoupling the hardware from the upper layer, it also helps make innovation faster and reduce the go-to-market time.
- The platform layer includes the operating system and application frameworks (e.g., Java framework) and other system components (e.g., data base and file system). Many popular cloud services operate at this level. For example, Microsoft Azure [14], Google AppEngine [15], and Amazon S3 [16] offer APIs for implementing typical web services.
- The Software-as-a-Service (SaaS) model means that the provider offers software on the common platform as well as the underlying database. Many traditional software companies (e.g., IBM, Microsoft, and Oracle) and the new players (e.g., Salesforce) are moving into this category. Cloud applications can automatically scale as the demand changes.
The layering architecture of cloud computing provides more modular design compared to traditional compute model. Resources are drawn whenever is needed on demand to fulfill a specific task. Unneeded resources can be relinquished, and the allocated resource is revoked after the job is done. Depending on the business model, cloud can be categorized to be private cloud where the data and processes are managed within the organization; public cloud where the resources and applications are provided/managed by a third-party off-site provider; and hybrid cloud where both internal and external cloud providers exist.
Figure 2.1 Cloud computing architecture.
2.1.2 Types of Clouds
Originally synonymous with public clouds, today cloud computing breaks down into three primary forms: public, private, and hybrid clouds.1 Each type has its own use cases and comes with its advantages and disadvantages.
Public cloud is the most recognizable form of cloud computing to many consumers. In a public cloud, resources are provided as a service in a virtualized environment, constructed using a pool of shared physical resources, and accessible over the Internet, typically on a pay-as-you-use model. These clouds are more suited to companies that need to test and develop application code and bring a service to market quickly, need incremental capacity, have less regulatory hurdles to overcome, are doing collaborative projects, or are looking to outsource part of their IT requirements. Despite their proliferation, a number of concerns have arisen about public clouds, including security, privacy, and interoperability. What is more, when internal computing resources are already available, exclusive use of public clouds means wasting prior investments. For these reasons, private and hybrid clouds have emerged, to make the environments secure and affordable.
Private clouds, can be defined in contrast to public clouds. While a public cloud provides services to multiple clients, using a shared infrastructure, a private cloud, as the name suggests, ring-fence the pool of resources, creating a distinct cloud platform that can be accessed only by a single organization. Hence, in a private cloud, services and infrastructure are maintained on a private network. Private clouds offer the highest level of security and control. On the other hand, they require the organization to purchase and maintain its own infrastructure and software, which reduces the cost efficiency. Besides, they require a high level of engagement from both management and IT departments to virtualize the business environment. Such a cloud is suited to businesses that have highly critical applications, must comply with strict regulations, or must conform to strict security and data privacy issues.
A hybrid cloud comprises both private and public cloud services. Hence, it is suited to companies that want the ability to move between them to get the best of both worlds. For example, an organization may run applications primarily on a private cloud but rely on a public cloud to accommodate spikes in usage. Likewise, an organization can maximize efficiency by employing public cloud services for nonsensitive operations while relying on a private cloud only when it is necessary. Meanwhile, they need to ensure that all platforms are seamlessly integrated. Hybrid clouds are particularly well suited for E-commerce since their sites must respond to fluctuating traffic on a daily and seasonal basis. On the downside, the organization has to keep track of multiple different security platforms and ensure that they can communicate with each other. Regardless of its drawbacks, the hybrid cloud appears to be the best option for many organizations.
In Table 2.1, we enlist the main benefits and risks associated with each type of clouds. Understandably, security is one of the main issues in cloud computing. There are many obstacles as well as opportunities for cloud computing. Availability and security are among the main concerns [19] [20].
Table 2.1 Cloud computing: benefits and risks
Cloud type Benefits Drawbacks Public- Low investment in the short run (pay-as-you-use)
- Highly scalable
- Quicker service to market
- Security: multitenancy and transfers over the Internet [18]
- Privacy and reliability [18]
- More control and reliability
- Higher security
- Higher performance
- Higher cost: heavy invest in hardware, administration and maintenance
- Must comply with strict regulations
- Operational flexibility: can leverage both public and private cloud
- Scalability: run bursty workloads on the public cloud
- Cost-effective
- Security, privacy and integrity concerns
2.1.3 Challenges
Being a disruptive technology, cloud computing has gained significant momentum in the past decade. However, it still faces several challenges with regard to performance, security, privacy, and interoperability. In the following, we discuss the challenges from these aspects:
- Guaranteed performance: Compared to the dedicated resource allocation in traditional compute model, cloud computing dynamically allocates resources on demand. While this is a key feature that enables multiplexing and elasticity, it also introduces serious concern on the application perceived performance. To provide guaranteed performance, not only the compute resources (e.g., CPU and memory) should be allocated sufficiently, but also the networking resources (e.g., bandwidth and low latency) should be satisfied. Many existing cloud platform uses virtualization to provide isolated compute resources. However, the networking is still shared in a best effort manner across all the tenants, which can become a severe bottleneck when the network is congested. In addition, sharing the same physical server may also introduce additional delay due to context switching. The provider needs to have ways to monitor applications' performance so that the service-level agreement (SLA) can be met.
- Security and privacy: Data...
