Results (
Vietnamese) 2:
[Copy]Copied!
Network Virtualization:
State of the Art and Research Challenges
ABSTRACT
Recently network virtualization has been pushed forward by its proponents as a long-term solution to the gradual ossification problem faced by the existing Internet and proposed to be an integral part of thenext-generation net- working paradigm. By allowing multiple hetero- geneous network architectures to cohabit on a shared physical substrate, network virtualization provides flexibility, promotes diversity, and promises security and increased manageability. However, many technical issues stand in the way of its successful realization. This article investi- gates the past and the state of the art in network virtualization along with the future challenges that must be addressed to realize a viable net- work virtualization environment.
INTRODUCTION
In recent years, the concept of network virtual- ization has attracted significant attention in the debate on how to model the next-generation net- working paradigm that can replace the existing Internet. Architectural purists view network vir- tualization as a tool for evaluating new architec- tures, whereas pluralists conceive virtualization as a fundamental diversifying attribute of thenext-generation architecture itself [1]. They believe that network virtualization can eradicate theso-called ossifying forces of the current Inter- net by introducing disruptive technologies [1, 2].
Network virtualization is defined by decou- pling the roles of the traditional Internet service providers (ISPs) into two independent entities [2, 3]: infrastructure providers (InPs), who man- age the physical infrastructure, and service pro- viders (SPs), who create virtual networks (VNs) by aggregating resources from multiple InPs and offerend-to-end services. Such an environment will proliferate deployment of coexisting hetero- geneous network architectures free of the inher- ent limitations of the existing Internet.
In this article we survey the past and the state of the art of network virtualization, and provide a better understanding of the key research chal- lenges. The rest of this article is organized as follows. First, four somewhat similar ideas (virtu- al local area networks [VLANs], virtual private networks [VPNs], programmable networks, and overlay networks) are briefly reviewed. Next, a reference business model and a conceptual archi- tecture of a network virtualization environment (NVE) are presented, identifying the character- istics and critical design factors to materialize it. Following this, a number of past and present research projects on network virtualization and related concepts are summarized. Finally, a detailed study of the key issues is presented emphasizing open research challenges with an objective to stoke wide interest among researchers in this field.
HISTORICAL PERSPECTIVE
The concept of multiple coexisting logical net- works has appeared in the networking literature several times in the past, and can be categorized into four main classes: VLANs, VPNs, active and programmable networks, and overlay net- works.
VIRTUAL LOCAL AREA NETWORK
A VLAN is a group of logically networked hosts with a single broadcast domain regardless of their physical connectivity. All frames in a VLAN bear a VLAN ID in the medium access control (MAC) header, and VLAN-enabled switches use both the destination MAC address and VLAN ID to forward frames. Since VLANs are based on logical instead of physical connections, net- work administration, management, and reconfig- uration of VLANs are simpler than in their physical counterparts. In addition, VLANs pro- vide elevated levels of isolation.
VIRTUAL PRIVATE NETWORK
A VPN is a dedicated network connecting multi- ple sites using private and secured tunnels over shared or public communication networks like the Internet. In most cases, VPNs connect geo- graphically distributed sites of a single corporate enterprise. Each VPN site contains one or more customer edge (CE) devices that are attached to one or more provider edge (PE) routers.
Based on the protocols used in the data plane, VPNs can be classified into the following broad categories.
Layer 1 VPN — The layer 1 VPN (L1VPN) framework emerged in recent years from the need to extend layer 2/3 (L2/L3) packet switch- ing VPN concepts to advanced circuit switching domains. It provides a multiservice backbone where customers can offer their own services, whose payloads can be of any layer (e.g., asyn- chronous transfer mode [ATM] and IP). This ensures that each service network has an inde- pendent address space, an independent L1 resource view, separate policies, and complete isolation from other VPNs.
Layer 2 VPN — Layer 2 VPNs (L2VPNs) trans- port L2 (typically Ethernet) frames between par- ticipating sites. The advantage is that they are agnostic about the higher-level protocols, and consequently more flexible than L3VPN. On the downside, there is no control plane to manage reachability across the VPN.
Layer 3 VPN — A layer 3 VPN (L3VPN) is characterized by its use of L3 protocols in the VPN backbone to carry data between the dis- tributed CEs. There are two types of L3VPNs.
In the CE-based VPN approach, the provider network is completely unaware of the existence of a VPN. CE devices create, manage, and tear down the tunnels between themselves. Sender CE devices encapsulate the passenger packets and route them into carrier networks; when these encapsulated packets reach the end of the tunnel (i.e., receiver CE devices), they are extracted, and actual packets are injected into receiver networks.
In the PE-based approach, the provider net- work is responsible for VPN configuration and management. A connected CE device may behave as if it were connected to a private net- work.
Higher-Layer VPNs — VPNs using higher-layer(e.g., transport, session, or application) protocols also exist. SSL/TLS-based VPNs are popular for their inherent advantages in firewall and NAT traversals from remote locations. Such VPNs are lightweight, easy to install and use, and provide higher granularity of control to their users.
Being translated, please wait..
