Introduction to new optical Internet technology based on other models

Traditional point-to-point WDM optical networks cannot provide the network scalability, low cost per bit, reserved rate, and ease of operation required to carry next-generation Internet services. To join the Internet-driven new economic competition, network operators must actively adopt new technologies and methods to establish a basic optical transport network that can support the growth of Internet businesses to expand and improve network performance.

Based on IPoverWDM technology, the Internet is an optimized IP optical network that runs directly on the optical network. It is a data communication network composed of high-performance WDM devices, gibits, and bitwise route exchange nodes. The optical Internet uses IP technology and WDM Technology to form a new high-speed broadband optical network technology.

1. Peering model optical Internet Topology

Over the past two years, through the efforts of the optical communication and industry sectors, the optical control plane under a common Multi-Protocol Label Exchange (GMPLS, the concept has quickly escalated from simple to a detailed set of protocol standards. In many cases, the optical control plane is an innovative technology that potentially revolutionizes the underlying transport infrastructure.

Peermodel is a network structure supported by the Internet Engineering Task Group (IETF). The emergence of universal Multi-Protocol Label exchange technology is an innovative network technology, it greatly promotes the development of Internet technology. GMPLS unifies the signaling and path establishment of various control planes, and exchange devices at each layer use the same signaling to control the user plane. To ensure that GMPLS signaling can be securely and reliably transmitted in various network topologies, the basic operations of the signaling system are request actions, connection-related attributes, protocols for transmitting operation commands over the network, and channels for transmitting signaling messages.

GMPLS signaling consists of the functional description of signaling (GMPLS-SIG), extended Resource Reservation protocol RSVP-TEresourcereservationsetupprotocol-traffic engineering) and extended restricted routing label distribution protocol CR-LDPconstraintbased routing label distribution protocol). GMPLS uses the parameters of the LSP, label switching path in the signaling switch label exchange channel, such as bandwidth, signal type, protection used, and location in special multiplexing ), and Bind tags on the channel using the RSVP-TE and CR-LDP protocols. Add non-group interface features to the "tag request message" in GMPLS. The tag allocation can be request-driven, data/stream-driven, or topology-driven. In addition, GMPLS also extends the MPLS routing protocol, defining two extended internal gateway protocols (IGP), OSPF-TE and IS-IS-TE ); it can broadcast links to various types of links (grouping, time slot, wavelength, and Optical Fiber Links) and support adjacent forwarding. GMPLS uses the constrained routing mechanism to allocate relevant transmission network topology information, including using IGP to expand and forward the status information of adjacent nodes.

1.2 characteristics of Peer-to-Peer Model

The Peer-to-Peer Model intelligently transfers the control of the optical transport layer to the IP layer for end-to-end control. The optical transmission network and IP network can be considered as a unified network. The optical switch and label exchange router have a unified routing area, and all information can be freely exchanged between them, and run the same routing and signaling protocols to achieve integrated management and Traffic Engineering and eliminate barriers between different network areas. A unified control plane can eliminate the complexity of managing a hybrid optical interconnection system. The control and operation semantics of this network are often separated and different.

The Peer-to-Peer Model breaks through the obvious boundaries between the transmission platform and the service layer. The two-layer devices have a peering relationship with each other, that is, IP Routers and optical cross-connections (OXC) devices are equivalent to each other, A common IGP protocol, such as a OSPF-TE or IS-TE, runs the same routing protocol in the optical and IP domains to exchange topology information. In the peer-to-peer model, all IP Routers and OXC devices share a common addressing scheme, and the service provider (rather than the optical core network) controls the use of the optical core network. Service providers can see the structure of the core optical network, so they can make optimal routing decisions.

2. Link Status Routing Protocol

2.1OSPF route Status Protocol and layered structure

The Open Shortest Path First Protocol (OSPF) is one of the internal gateway protocols. It is generated because the routing information protocol (RIP, routinginformationprotocol) is not applicable to the rapidly changing network environment. Compared with RIP, there is no need for hop in the OSPF network, and its convergence speed is much faster than RIP. OSPF has three types of specifications (metric) and can be combined to facilitate network load balancing. OSPF only broadcasts the changed link state information to the entire network when the network status changes, which can save network bandwidth, which is especially important for WAN. OSPF supports a variable-length Subnet Mask (VLSM). Different ports on a vro can have different subnet masks, making the allocation of network IP addresses more flexible. Because OSPF has the above advantages, it is widely used in the selection of routing protocols.

OSPF is a typical link status routing protocol, which is generally used in the same routing domain. The routing domain here refers to an autonomous system (AS, autonomoussystem), which is a network that exchanges route information through a unified routing policy (RP, routingpolicy) or routing protocol.

OSPF is layered in an AS. AS can be divided into backbonearea and several regions, each of which is relatively independent. Vrouters in the same region adopt the same process and algorithm, and the link status database is the same. Each region has a unique Region ID, through the edge router (BR, borderrouter)/Regional edge router (ABR, area border router.

Based on its functions and locations in the network topology, OSPF routers can be roughly divided into the following four types:

(1) internal router (IR, internalrouter): All router interfaces are in the same region, and all the internal routers in the same region have the same link status database.

(2) edge router: at least one interface is connected to the master area 0.

(3) Regional edge router: the router connected to multiple regions. These routers maintain independent link status databases for each connected region. The ABR is the exit point of the region, that is, the traffic in the region must pass through the ABR to reach other regions. The ABR summarizes the link status database of the connected region ). One region can have one or more ABR.

(4) Autonomous System edge router (ASBR, autonomoussystemboundaryrouter): at least one interface is connected to an external network (other AS, such AS a non-OSPF network ). These routers can communicate information about OSPF and non-OSPF networks.

A vro can act as a vro of different types and concurrently serves multiple tasks.

We can see that area 1, Area 2, and area 3 are connected to the master Region 0 through their respective Edge Routers. The master Region 0 can be viewed as a special region and the Region ID is 0. Other regions must be physically connected to the master Region 0. Of course, if a region N cannot be directly physically connected to the trunk region, you can use a virtual link to connect to the trunk region 0 across another region.

In this AS, all OSPF routers maintain the same information database (IDB), which describes the AS structure. The database stores the status information of the corresponding link in the routing domain. The OSPF router calculates the OSPF route table through IDB. As a routing protocol for link status, OSPF transmits link status broadcast (LSA, linkstateadvertisement) packets to all routers in a region, which is different from distance vector routing protocol (RIP. A router that runs the distance vector routing protocol (RIP) Transmits some or all electric meters to its adjacent routers. All interface information, all measurements, and other variables are included in the OSPF link status broadcast. A router that uses OSPF must first collect the relevant link status information and calculate the shortest path to each node based on certain algorithms, the routing protocol that provides distance vectors only sends route update information to its adjacent routers. Different from RIP, OSPF divides an autonomous system into several regions. There are two routing Methods: when the source and destination are in the same region, intra-region routing is used; when the source and destination are in different regions, the inter-region routing is used. This greatly reduces network overhead and increases Network stability. When a vro in a region fails, the normal operation of vrouters in other regions of the autonomous system is not affected, which facilitates network management and maintenance.

2.2IS-IS Routing Protocol

The routing protocol between the intermediate system and the intermediate system (IS-IS) IS the OSI Model Representation Method of the router. It IS used for TCP/IP-based IP networks. IS-IS can be easily expanded, mainly IPv6. The IS-IS system IS divided into two layers: Bone Layer 2 (L2) and Region Layer 1 (L1). A router can only belong to one region. The Ll router only knows the topology in the current region. All traffic destined for other regions is sent to the nearest L2 router. The L2 router must form a trunk, which is similar to the OSPF trunk region 0.

3 optical Internet routing mechanism in peer-to-peer Model

In the peer-to-peer model, the relationship between IP/GMPLS and the optical layer is equivalent. A unified control plane runs on the IP/GMPLS and optical layer at the same time, therefore, the control plane regards OXC devices as another type of routers, and the routers and OXC devices perform full topology information interaction. The management domain of the unified control plane includes both core optical network devices and edge network devices. This allows Internet service providers (ISPs) edge network devices to understand the topology of the core network, and participate in route computing.

When only one optical network IS involved, General IGP (such as OSPF or IS-IS) can be appropriately expanded to publish topology information on the entire optical Internet. For OSPF, non-transparent LSAs can be used to broadcast topology status information. The OSPF protocol establishes and maintains the adjacent relationship through the Hello protocol data packet, and uses it to ensure two-way communication between adjacent routers. The OSPF router periodically sends Hello data packets. When the router sees its own Hello data packets listed in other routers, two-way communication is established between the two routers. In a multi-access environment, Hello data packets are also used to discover the specified router (DR) and control which routers are used to establish interaction with them through DR.

The second step after two OSPF routers establish two-way communication is to synchronize databases. Database Synchronization is the most common feature of all link status routing protocols. In the OSPF routing protocol, the Database Synchronization relationship is only maintained between routers that establish an interactive relationship. The OSPF database is synchronized through the OSPF database description packet (datadescriptionpackets. The OSPF router periodically generates database description data packets (the data packets are ordered with serial numbers) and broadcasts these data packets to neighboring routers. The adjacent router can compare the serial number of the data packet according to the database description with the data in its own database. if it finds that the received data is larger than the data serial number in the database, A request is sent for data with a large serial number and the data obtained from the request is used to update the link status database. The process from sending Hello data packets and establishing Database Synchronization to establishing full OSPF interaction between OSPF routers can be divided into several different states.

For IS-IS, an appropriate extension IS required for IP routing. When an optical Internet contains multiple domains, Inter-Domain Routing and signaling are required. However, in any case, a general addressing mechanism is required for Optical Networks and IP networks, A common address space can be achieved by using IP addresses in both the IP and optical fields. In this way, the network elements in the optical network become addressable entities in the IP network.

The routing mechanism based on the peer-to-peer model is integrated routing. In this way, IP domains and optical domains run the same IP routing protocol, such as OSPF for proper optical domain extension, which must contain Optical Link parameters and specific limits of optical networks. All nodes (OXC devices and routers) in the network store the same topology and link status information, so that the router can calculate the end-to-end route to the other router in the optical network, such a label exchange channel can be established with GMPLS signaling (such as RSVP-TE or CR-LDP ). When LSP is routed over the optical network, an optical channel must be established between two edge routers. This optical channel is essentially a tunnel in the optical network. Its capacity is larger than that of the first LSP. Other routers in the network can route other LSP in this optical channel to make full use of resources. Therefore, this optical channel can be advertised as a virtual link in the topology.

Forward neighbor (FA, forwardingadjacency) is an important concept. It is very important to spread existing optical channel information to other routers. FA is essentially a virtual link advertised according to the link routing protocol. It can be described with the same parameters as conventional Link resources. Although it is necessary to explain the establishment mechanism of FA, it does not need to explain how FA is applied in the routing mechanism. Once FA is advertised in the Routing Status Protocol, its use will also be defined in the routing computing and traffic engineering method.

The peer-to-peer model can seamlessly interconnect an IP network with an optical network, provided that the routing information of the optical network must be understood by the IP router network.

Peer-to-Peer Model is a new type of network technology and a network structure supported by IETF. Therefore, IETF proposes the concept of universal multi-protocol tag exchange. The basic idea is to use the IP layer for MPLS channel routing and signaling. After some modifications, it is directly used for Connection Control at various layers, including the optical transport layer. The emergence of GMPLS technology makes the traditional multi-layer network structure between IP address and WDM flat, this is a key step for the transformation of the transmission network from circuit switching to group switching, and the combination of optical network layer transmission and switching functions.

The routing protocol is a key technology in the construction of the optical Internet. It is usually used for the OSI reference model of the layer 3rd (that is, the network layer) and TCP/IP Internet. Depending on the path determination method, the routing protocol can be divided into link status routing protocol, distance vector routing protocol and Hybrid Routing Protocol. OSPF is an internal routing protocol designed for IP networks and is suitable for large and variable networks. OSPF features fast convergence, high update efficiency, and no hop limit. It supports a variable-length Subnet Mask (VLSM). It selects a path based on the bandwidth. Studying the routing mechanism under the peer-to-peer model is of great significance for the establishment of the optical Internet.