Results (
Vietnamese) 2:
[Copy]Copied!
ACTIVE AND PROGRAMMABLE NETWORKS
Active and programmable networks research was motivated by the need to create, deploy, and manage novel services on the fly in response to user demands. In addition to programmability, they also promote concepts of isolated environ- ments to allow multiple parties to run possibly conflicting codes on the same network elements without causing network instability.
Two separate schools of thought emerged on how to actually implement such concepts.
The Open Signaling Approach — Open sig- naling takes a telecommunication approach with a clear distinction between transport, control, and management planes that constitute pro- grammable networks, and emphasizes quality of service (QoS) guarantees. An abstraction layer is proposed for physical network devices to act as distributed computing environments with well defined open programming interfaces allowing service providers to manipulate network states.
The Active Networks Approach — Active networks promote dynamic deployment of new services at runtime within the confinement of existing networks. Routers or switches in these networks can perform customized computa- tions based on the contents of the active pack- ets and can also modify them. Active networks allow the customization of network services at packet transport granularity and offer more flexibility than the open signaling approach at the expense of a more complex programming model.
OVERLAY NETWORKS
An overlay network is a logical network built on top of one or more existing physical networks. The Internet itself started off as an overlay on top of the telecommunication network. Overlays in the existing Internet are typically implement- ed in the application layer; however, various implementations at lower layers of the network stack do exist.
Overlays do not require or cause any changes to the underlying network. Consequently, over- lays have long been used as relatively easy and inexpensive means to deploy new features and fixes in the Internet. A multitude of application layer overlay designs have been proposed in recent years to address diverse issues, which include ensuring performance and availability of Internet routing, enabling multicasting, provid- ing QoS guarantees, protecting from denial of service attacks, and content distribution and file sharing services. Overlays have also been used as testbeds (e.g., PlanetLab) to design and evaluate new architectures.
The authors in [1] point out that standard overlays falter as a deployment path for radical architectural innovations in at least two ways. First, overlays have largely been used as a means to deploy narrow fixes to specific problems with- out any holistic view. Second, most overlays have been designed in the application layer on top of IP; hence, they cannot go beyond the inherent limitations of the existing Internet.
NETWORK VIRTUALIZATION
ENVIRONMENT
Unlike the existing all-IP Internet, a virtualized networking environment is a collection of multi- ple heterogeneous network architectures from different SPs. Each SP leases resources from one or more InPs to create VNs, and deploys cus- tomized protocols and services.
BUSINESS MODEL
The main distinction between the participants in the network virtualization model and the tradi- tional model is the presence of two different roles, InPs and SPs, as opposed to the single role of the ISPs[2–4].
InP — InPs deploy and actually manage the underlying physical network resources. They offer their resources through programmable interfaces to different SPs. InPs distinguish themselves through the quality of resources they provide, the freedom they delegate to their cus- tomers, and the tools they provide to exploit that freedom.
Unlike the existingall-IP Internet, a virtualized networking environment is a collection of multiple heterogeneous network architectures from different SPs. Each SP leases resources from one or more InPs to create VNs and deploys customized protocols and services.
End users in the net- work virtualization model are similar to those of the existing Internet, except that the existence of mul- tiple VNs from com- peting SPs provides them a wider range of choice. Any end user can connect to multiple VNs from different SPs for dif- ferent services.
SP — SPs lease resources from multiple InPs to create and deploy VNs by programming allocat- ed network resources to offer end-to-end ser- vices to end users. An SP can also provide network services to other SPs. It can also create child VNs by partitioning its resources and act as a virtual InP by leasing those child networks to other SPs (Fig. 1).
End User — End users in the network virtualiza- tion model are similar to those of the existing Internet, except that the existence of multiple VNs from competing SPs provides them a wider range of choice. Any end user can connect to multiple VNs from different SPs for different services.
ARCHITECTURE
In an NVE the basic entity is a VN. A VN is a collection of virtual nodes connected together by a set of virtual links to form a virtual topology, which is essentially a subset of the underlying physical topology. Each virtual node is hosted on a particular physical node, whereas a virtual link spans over a path in the physical network and includes a portion of the network resources along the path.
Each VN is operated and managed by a sin- gle SP, even though the underlying physical resources might be aggregated from multiple InPs. Figure 1 depicts two VNs, VN1 and VN2, created by service providers SP1 and SP2, respectively. SP1 composed VN1 on top of the physical resources managed by two different InPs (InP1 and InP2), and providesend-to-end services to end users U2 and U3. SP2, on the other hand, deployed VN2 by combining resources from infrastructure provider InP1 with a child VN from service provider SP1. End users U1 and U3 are connected through VN2.
The owner of a VN is free to implement end-to-end services by deploying custom packet for-mats, routing protocols, forwarding mechanisms, as well as control and management planes. As men- tioned earlier, end users have the choice to opt in to any VN. For example, U3 is subscribed to VN1 and VN2 managed by SP1 and SP2, respectively.
ARCHITECTURAL PRINCIPLES
Network virtualization propounds the following principles for the next-generation networking paradigm.
Coexistence — Coexistence of multiple VNs is the defining characteristic of an NVE [1–3]. It refers to the fact that multiple VNs from differ- ent SPs can coexist together, spanning over part or full of the underlying physical networks pro- vided by one or more InPs. In Fig. 1, VN1 and VN2 are two coexisting VNs.
Recursion — When one or more VNs are spawned from another VN creating a VN hierar- chy withparent-child relationships, it is known as recursion as well as nesting of VNs [5]. Service provider SP1 in Fig. 1 leased away a portion of its allocated resources to SP2, to whom it appears simply as a virtual InP.
Inheritance — Child VNs in an NVE can inher- it architectural attributes from their parents, which also means that the constraints on the parent VN automatically translate to similar constraints on its children [5]. For example, con- straints imposed by InP2 will automatically be transferred to VN2 from VN1 through inheri- tance. Inheritance allows an SP to add value to the spawned child VNs before reselling them to other SPs [3].
Revisitation — Revisitation [6] allows a physi- cal node to host multiple virtual nodes of a sin- gle VN. Use of multiple logical routers to handle diverse functionalities in a large complex net- work allows an SP to logically rearrange its net- work structure and to simplify the management of a VN. Revisitation can also be useful for cre- ating testbed networks. Figure 1 provides an example of revisitation in VN2.
Being translated, please wait..
