Integrating IP and Optical Networks (Transport Area)


In building networks today, a circuit or path setup requires touching multiple transport networks and technologies. For example, in creating an IP connection of OC-12 bandwidth from New York to San Francisco, a Layer 2 circuit must be available, in addition to the IP equipment and connectivity. To provision that Layer 2 circuit, you might have to provision optical connections on the Synchronous Optical Network/Synchronous Digital Hierarchy (SONET/SDH) network and provision Dense Wavelength Division Multiplexing (DWDM) long-haul for this OC-12 connection.

Each specific technology has its own control protocol and, as a result, each set of control protocols does not communicate directly with the others at a peer level. Instead, as we see in our example, networks are layered one on top of the other creating overlays at each layer to collectively provide end-user servicesDWDM long-haul that connects multiple OC-192s via wavelengths, a SONET/SDH Optical Add and Drop Multiplexer (OADM) that multiplexes OC-12s to OC-192s, and an MPLS LSP that provides IP layer connectivity. Obviously, this process requires knowledge of each technology domain, provisioning of each layer, and separate management of per-domain operation functions.

A unified control plane with a common set of control functions that can tie in all transport types and provision across all technologies can make the provisioning process simple and efficient. The use of such a control plane enables quick deployment of IP services and applications, and providers no longer need to separately provision connectivity across technologies. The common control plane signals appropriate parameters across different types of transport networks. The common control plane uses IP-like routing and signaling and allows topology discovery and connection setup, as well as a sharing of resource and state information across different technology domains.

How Does it Work?

Although all the details of GMPLS and Unified Control Plane (UCP) are beyond the scope of this book, we briefly describe how this works to understand how the future deployment of MPLS and GMPLS is shaping up.

To understand the basic concepts of GMPLS, let us first understand the layering model. What is commonly represented in the network diagrams of connectivity between two routers via an OC-12 or OC-192 link is much more than that. The OC-192 from the router usually terminates onto an optical add-drop MUX. Multiple add-drop MUXes aggregate to DWDM equipment that are connected by long-haul fiber. In other words, in the layering concept, at the lowest layer are the devices that connect long-haul fibera DWDM network. Layered on top of this are the OADM devices that provide the SONET framing optical grooming and Optical Carrier Level n (OCn) connectivity to the routers. (See Figure 15-1.)

Figure 15-1. GMPLSLayered Model


A common control plane network that connects the IP routers, SONET/SDH MUXes, optical cross connects, and DWDM gear provides the signaling of parameters such as wavelengths, TDM channel numbers, and fiber ports in the GMPLS signaling. Here the TDM channel number or the wavelength lambda is the label. The signaling is done by extending RSVP-TE to carry optical network parameters, and admission control is performed based on available bandwidth (optical channel, fiber port, or light wavelength), as shown in Figure 15-2.

Figure 15-2. GMPLSOut of Band Control Plane


The optical light path setup is similar to an RSVP-TE tunnel setup where the head-end node signals using RSVP. The setup messages travel the network and initialize bandwidth (ports, optical channels, or wavelengths) to the tail end. However, the difference here is that signaling messages can either travel in band or out of band through a control network or control channel. In optical networks an out-of-band network commonly carries the provisioning and management information; sometimes a dedicated control channel provides this information in band. By using a common signaling protocol with appropriate technology, specific extensions, light paths, or TDM connections can be dynamically triggered when a demand exists for bandwidth at the IP layer.

Note

While writing this book, we recalled a conversation that happened with a service provider. We paraphrase this conversation to demonstrate the difficulty providers face in provisioning circuits between two cities or points and the management challenges they face.

In discussing the provisioning times of circuits in general, we were surprised to hear that it takes the provider between three weeks and three months to light up an OC-192. They create a service request and toss it over to the transport group. Depending on the geography, available capacity of fiber, free channels, cross-connect capacity, and so on, the transport group provisions this request and tosses the connections back to the IP group. Because these networks, transport networks, and IP networks are managed and supported by different groups within this provider, the IP group has no control over where the OC-192 gets provisioned and what its delay budget is. In fact, the IP group notices a change in reroute of this OC-192 by monitoring the variation in ping times of this link.

This shows the challenges faced by the provider and the difficulty they have in rapidly provisioning and managing circuits. If they had the ability to control the placement of this OC-192 (along which optical path) and the ability to provision it rapidly (inline with the demand of the IP capacity), their bottom line would be much better. Delays in availability of network capacity mean higher operations costs, which affect the bottom line.


Bandwidth On-Demand Service

In addition to providing a rapid provisioning of circuits, GMPLS can be used as a standard interface like a network-to-network interface (NNI) to provide a service-like bandwidth on demand to other regional or national providers or large enterprises.

Let us consider a facilities-based provider that has a transport network. This provider wants to offer bandwidth on-demand services to regional ISPs or large enterprise customers that need loads of bandwidth for bulk data transfers for relatively short periods of time. The transport provider needs the ability to rapidly or instantly provision circuits, depending on the demand. The transport provider has two choices.

First, if the optical and TDM equipment is capable of GMPLS control plane, then you enable the control plane, create an NNI with the regional providers, and react to signaling requests from the regional providers using the GMPLS network. The dynamically signaled requests from the customers are authenticated and honored by setting up the cross connections and allocating OCn channels from the ingress to the egress of the network.

The second choice is this: If the existing optical and TDM equipment is not capable of being upgraded with the unified control plane and GMPLS signaling, you might be able to create a proxy function that can translate the GMPLS requests coming from the regional ISPs or customers and translate this to the provisioning mechanisms of the equipment. A proxy function behaves similarly to the GMPLS peer and responds to signals in the same manner as a real GMPLS network. The proxy function translates the request to the existing provisioning systems to set up the OCn channel or light path hop by hop.

Due to the dynamic nature of this signaling, circuits can be signaled and torn down on demand based on customers's needs. Such a service can be attractive to service providers.

Challenges Faced with GMPLS and UCP

Any new protocol has its set of challenges with respect to operations and management. Moreover, GMPLS challenges the fundamental assumptions of network operations. Service providers have separate groups managing different network componentsone that manages the transport network and another that manages the IP network. Almost all ILECs, IXCs, and PTTs operate this way. Integrating the transport network with the IP network is a massive challenge for these providers for political and operational reasons. After the providers overcome their political and operational problems, GMPLS can help build an efficient network.




MPLS and Next-Generation Networks(c) Foundations for NGN and Enterprise Virtualization
MPLS and Next-Generation Networks: Foundations for NGN and Enterprise Virtualization
ISBN: 1587201208
EAN: 2147483647
Year: 2006
Pages: 162

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net