IP over SDH and IP over ATM Technology

1 Introduction
In recent years, with the exponential growth of the Internet, it has put forward higher requirements for WAN bandwidth. Internet service providers (ISPs) are actively exploring various technologies to better connect to their backbone routers. The common approach is the adoption of asynchronous transfer mode ATM) technology, using 155 M (STM-1) up to 622 M (STM-4) of high-speed links. As a result, many IP over ATM technologies are generated, such as LANE, CIPOA, and MPOA.
However, because the carrier's underlying network structure is fiber-connected Synchronous Digital System SDH) North America is synchronous fiber network SONET), this makes the ISP more inclined to IP over SDH technology, to improve bandwidth efficiency, rather than conventional IP over ATM technology. These two solutions have aroused a strong debate in the industry.
2 SDH/SONET Overview
SDH is a physical layer technology used for transmission and reuse. The transmission rate can reach 10 Gbps, which is an ITU Standard. SONET is the ANSI standard of the American Standards Committee. The two are only different in the reuse mechanism, while other technologies are similar. Therefore, this article takes IP over SDH as an example, the following describes SDH briefly.
2.1 SDH network elements
SDH network elements mainly include Synchronous Optical Fiber line system, terminal multiplexing TM), plug-in Multiplexing ADM) and Synchronous Digital crossover equipment DXC ). TM is mainly used to re-connect branch signals to STM signals and complete their photoelectric conversion and inverse processes. ADM has the flexible function of inserting and separating circuits, users generated by sources such as ATM switches can be inserted and split into SDH frames in the ring. DXC completes the signal crossover. A typical SDH application is a dual-ring optical fiber application. The dual-ring structure uses automatic protection switching to achieve double-ring self-healing.
2.2 SDH frame structure
SDH transmits 8 k sdh frame STM-N per second), STM-N frame is based on the frame structure of STM-1. Although SDH provides a synchronous frame structure, it does not force the user's net load to be located in a specific location in the SDH frame. On the contrary, it allows the user's net load to fluctuate within the frame, use the pointer in the overhead field to indicate the starting position of the user's net load. In the user's opinion, SDH is a physical layer medium that provides byte synchronization.
3 ATM Overview
ATM is a connection-oriented transmission mechanism based on CELL). A fixed 53-byte atm cell facilitates hardware-based exchange. The ATM uses a VCC or VPC connection and the VPI/VCI in the cell header to identify each connection.
Not only does data link layer technology provide complete network layer and transmission layer features, such as addressing, routing, and traffic control. ATM allows multiple user data streams to share valid link bandwidth, however, QoS must be set in advance for each connection. Voice, data, images, and video streams can all be applied to ATM. However, different application flows require different ATM Adaptation layers (AAL) to map user data to ATM cells.
An ATM can run on different physical media. The ATM layer generates cells and sends them to the physical layer to send and receive corresponding signals from the physical media. SDH/SONET is one of the physical layers of ATM. Because the Net Load in SDH frames is not an integer multiple of 53-byte cells, ATM cells can only be directly sent to the Net Load in SDH frames continuously. When receiving, the Cell Header error detection HEC in the ATM cell header is used to describe the correctness of the ATM Cell in the SDH net load.
4 IP over ATM
IETF has decided to use IP addresses to interconnect the networks and build a single, huge IP network. A single network usually uses different physical links and network technologies, but if the IP runs on various network layers, the network and network will achieve seamless interconnection. Currently, IP addresses are running on a variety of network technologies, including broadcast LAN technologies, such as Ethernet, circuit switching WAN technologies, such as X.25, and group switching WAN technologies, for example, SMDS.
In 1990, IETF began to define IP over ATM. The ATM Forum started to run different layer 2 and Layer 3 protocols on the ATM, especially in terms of LAN.
4.1 classic IP over ATM (CIPOA)
The objective of CIPOA is to use ATM as the low-layer data link layer of the IP address, while the application layer is based on the traditional IP address. Originally, in the traditional IP network, the implementation of ATM only replaces the LAN line with ATM. As a result, the ATM network needs to be divided into different logical subnets (LIS), and routers are required for communication between LIS. There is no Broadcast Function in the ATM network. Therefore, the traditional broadcast Address Resolution Protocol ARP is replaced by the customer/Server mode atm arp protocol.
A default Logical Link/subnet Access Protocol (LIC/SNAP) encapsulates 8-byte segments for transmitting IP addresses and atm arp packets over an ATM, these packages are encapsulated and adapted using AAL5 and then mapped directly to the ATM cells. These cells are transmitted using virtual connections to a pre-defined PVC or switched SVC. For SVC call establishment, the UNI3.1/4.0 of the ATM Forum or the Q.2931 signaling of the ITU-T is required.
4.2LANE LAN simulation
LANE is introduced by the ATM Forum to simulate Eerthnet/802.3 and Token Ring/802.45 on the ATM network. By using LAN simulation, existing LAN applications can communicate on the ATM network, just like on a traditional LAN that uses MAC addresses for addressing, it can provide multicast and broadcast data transmission. LANE runs on the MAC layer, and any layer-3 protocol can run on it. On the contrary, CIPOA can only run the IP protocol.
A simulation LANELAN is composed of a group of LAN simulation servers (LES), broadcast and unknown server BUS), and LAN simulation configuration server LECS. A simulated LAN is a lan cidr block. A bridge or router is required to communicate with other lan cidr blocks. To transmit LAN control and datagram, a 2-byte encapsulation group is used on the ATM. The new LANE version allows the LLC/SNAP encapsulation, And the LANE package is encapsulated into an ATM Cell After AAL5 adaptation, it is carried by SVC.
4.3 MPOA
MPOA is another protocol in the ATM Forum, which overcomes one of the main disadvantages of CIPOA and LANE. It refers to the intermediate router for communication between different subnets. The intermediate router needs to assemble the cells into a layer-3 package, after routing is selected, packets are encapsulated into ATM cells for forwarding. MPOA allows users in different subnets to directly establish a short VCC connection without intermediate restructuring and segmentation, in the same subnet, MPOA and LANE are the same.
MPOA includes the MPOA client and MPOA server. The MPOA client can be an ATM host or a non-ATM network segment host connected to the ATM through edge devices. The edge devices can perform layer-2 bridging or layer-3 forwarding between them, transfer through short-path VCC. The MPOA client monitors layer-3 grouped streams. When detecting continuous grouped streams destined for a specific destination, the MPOA client queries the destination ATM address or the ATM address of the ATM edge device to the destination to create a short path VCC. The MPOA server uses the next-Hop Resolution Protocol (NHRP) defined by IETF to route the resolution package to the egress ATM address of the network destination. MPOA uses the Distributed Virtual routing technology to connect the ATM subnet and edge devices of the traditional LAN subnet, similar to the interface card of the virtual router, the entire ATM network connected to the edge device is the forwarding backplane of the virtual router. The Group forwarding function is separated from the routing computing function. The routing computing function is completed by the Routing Server. This separation improves forwarding efficiency and throughput compared with traditional routers. The group uses the LANE or LLC/SNAP Encapsulation Format, and AAL5 is used to directly adapt to ATM cells, and SVC is used for connection.
4.4 protocol stack
CIPOA, LANE, and MPOA are executed as the common protocol layer PLOA on the ATM, including the Logical Link Control LLC for the LLC/SNAP encapsulation) module and interface signaling and ATM Adaptation Layer and host Call Control (HSCC) module for VCC establishment and data transmission. The protocol stack is used to describe different IP over ATM technologies mentioned above.
4.5 MPLS Multi-Protocol Label Switching
The MPLS defined by IETF is the integration model of ATM and IP, rather than the traditional overlapping model. In MPLS, each vro is a vswitch, and a fixed length Label is added before the header of the traditional third-layer, create a label route table similar to a traditional route table in each MPLS Router. When forwarding data reports, search for these label route tables with fixed lengths to quickly forward data packets. The traditional route table search adopts the longest prefix matching method. MPLS uses a dedicated label Allocation Protocol (LDP). This protocol is used to establish a correspondence between a traditional route table and a label route, A label-based datagram forwarding shortcut is created between MPLS routers. MPLS can also use Open Shortest Path First Protocol OSPF) to establish a shortcut path without using the LDP protocol.
MPLS currently only supports IPv4, but its applicability is not limited to ATM. Its goal is any data link layer technology, as long as it supports identifying fixed-length labels of shortcuts, such as Ethernet.
5 IP over SDH

Because SDH rings provide point-to-point connections between routes, IP packets must be mapped to point-to-point links. The most common technology is IETF's Point-to-Point Protocol (PPP) encapsulation. IP over SDH is also called Packet over SDHPOS) technology.

5.1PPP

PPP is a point-to-point link layer technology that provides the following functions: encapsulate and transmit different network layer protocol datagram on the same link layer, establish, configure, and test link layer connections, and establish and configure network layer protocols. PPP specifies the encapsulation mechanism and Link Control Protocol LCP). In addition, PPP also requires authentication, link quality monitoring, and network control protocol NCP ).

PPP adopts a frame mechanism like HDLC and can run on various physical interfaces, such as RS232, RS422, and V.35.

5.2PPP over SDH

PPP regards SDH as a byte synchronization link, and uses PPP frames as byte streams to map to the SDH Net Load. In HDLC-like frames, corresponding signs are used to identify PPP frames in SDH net load.

5.3 protocol stack

5.4POS backbone technology

IP over SDH is used in the backbone network. The CPE device connects to the access server directly, and the Access Server is connected to the backbone router. The backbone router connects to the SDH double ring to form the backbone network.

6 Comparison of IP over ATM and IP over SDH Technologies

IP over SDH and IP over ATM are two main technologies, which have the following differences:

6.1 Protocol overhead

The maximum reason for ISP to enable IP over SDH is that the overhead of the ATM cell header is 5/53). Additional overhead include AAL5 padding and 8-byte tail) and LLC/SNAP encapsulate 8 bytes ). For IP over ATM, assuming the physical layer uses the SDH STM-1 link, the IP packet size is 576 bytes.

For an IP over PPP over SDH, it is also a STM-1 link with an IP packet size of 576 bytes.

The above comparison shows that the IP address running on the ATM only uses nearly 80% of the line rate, while the IP address running on the PPP over SDH is 95%. For the WAN main router, the extra IP over SDH capacity will be very attractive, because in the Wide Area backbone network, the bandwidth is relatively expensive, rather than the local area network or campus network bandwidth is relatively cheap.

6.2 bandwidth management

An ATM provides a full set of bandwidth management functions to manage and assign to each information stream VCC. It allocates bandwidth to these VCC instances based on the required QoS. Due to the nature of Cell Exchange, ATM allows multi-information stream sharing links and ensures a certain bandwidth for each information stream to satisfy its QoS.

PPP does not provide any bandwidth management functions. It simply provides point-to-point links. IP addresses must schedule their group forwarding functions to ensure that each information stream has a fair share of the link bandwidth. This will cause problems on low-speed links. When a large package with a low priority is transmitted, other high-priority packages will be blocked, for example, transferring a low-priority File Transfer packet will block the transmission of latency-sensitive voice packets. For real-time applications with latency 3, such latency changes on bandwidth-restricted links offset the bandwidth efficiency advantage brought by IP over SDH.

6.3 QoS

Parameters related to service quality include end-to-end latency, latency jitter, discard rate, and throughput. ATM provides a wide range of QoS parameters to negotiate with each VCC. The intelligent queuing and scheduling mechanism ensures the negotiated QoS parameters. ATM provides different service types to be used in coordination with different requirements. For example, applications with high QoS requirements can use CBR and VBR services, but applications with low QoS requirements, you can use the ABR and UBR services. The attributes of these ATMs make it easy to provide QoS at the IP level. information flows with certain QoS can be mapped to VCC with corresponding QoS, for example, voice streams are mapped to CBR or VBR connections, while file transfers can be mapped to ABR connections.

PPP runs on a single point-to-point link and does not provide any QoS capabilities. The IP layer must manage its packet transmission to ensure the service quality of the corresponding information flow.

Although ATM provides a wide range of QoS parameters, QoS-based services are limited to the ATM links between connected routers. To provide end-to-end QoS IP addresses, the router must provide intelligent queuing and Scheduling Mechanisms. Therefore, when the IP network is superimposed on the ATM network, although the actual communication occurs between the switches on the ATM network, but for the router, all of them regard ATM connections as point-to-point links similar to PPP.

6.4 addressing and Routing

ATM is defined as a complete network technology and has a huge function of terminal system addressing and routing connection. The ATM network can span geographical areas regardless of the router location, A wide range of connection mechanisms can be provided for them. On the contrary, PPP runs only on direct point-to-point links without addressing and routing capabilities.

When an IP over ATM is running, you need to translate the IP address into the corresponding ATM address to implement routing. This increases the complexity of the Protocol, the traffic required for address translation increases the overhead of the network. PPP does not require address translation, so there is no additional overhead.

6.5 network scalability

In the Backbone Router network, most routers need to communicate with each other, which means a mesh connection is required, whether it is PPP or atm svc.

In the IP over SDH Backbone Network, the backbone router must provide a point-to-point link. To prevent link faults, or even establish multiple links, in some scenarios, in order to reduce the number of hops of the router, the link between the backbone routers must be configured in a full mesh. The link required for the full mesh network is N * (N-1)/2, that is, the so-called N2 increase. An all-mesh network is expensive and cannot be expanded due to limited SDH links in a wide area.

In IP over ATM, because SVC is used, ATM can provide any connection between routers without the need for a full mesh network, or even when the ATM link fails, A dynamic SVC route can be used to find a roundabout route to ensure the connection between any two routers. Another advantage is that one ATM interface can be used to interconnect with multiple routers, so as to achieve more flexible network engineering and strong scalability, it can route SVC on different links and connect a vro to multiple different sinks using the same ATM link to solve the N2 problem.

6.6 Traffic Control

ATM uses Connection Admission Control CAC), traffic shaping, and usage Parameter Control UPC) or policies to ensure that the information flow complies with the traffic contract. The excess part will be marked and discarded when the network is overloaded, end users get congestion information through tagged or discarded packets. The interoperability between ATM Cell-level discard and TCP packet discard is very poor. to eliminate this phenomenon, many technologies have been introduced in the industry, such as partial packet discard PPD) and early packet discard EPD, to make the ATM recognition packet AAL frame) boundary, so that the entire packet is discarded when the network is overloaded, rather than ATM cells.

The ATM Forum also defines the ABR Service, which provides explicit feedback traffic control and uses the allowed rate to indicate the rate at which users can send data to the network, this rate will change as the network load changes, allowing users to access valid bandwidth without causing network overload. In the ideal state, the ABR service will eliminate the cell discard rate and push the network congestion conditions to the edge of the ATM network, this also requires the vro to cache more packets to adapt to the changes in the access rate allowed by the ATM network.

PPP does not provide any traffic control mechanism. Only TCP traffic control runs directly on the PPP link. As mentioned above, whether using ATM or using SDH directly, routers regard it as a channel with a certain bandwidth between two points. corresponding buffer mechanisms must be used to ensure reasonable throughput.

6.7 multi-protocol encapsulation

ATM provides two mechanisms to allow multiple protocols to share the same ATM link. A vcc is reused, and each protocol is allocated with a separate VCC. The ATM layer multiplexing and VCC are used. You do not need to add any other encapsulation headers to distinguish different protocols. Another mechanism is LLC multiplexing, which allows multiple protocols to reuse the same VCC. It adds an 8-byte Encapsulation Header to each packet to identify which Protocol it belongs, this multiplexing technology can be used in scenarios where the number of VCC instances is limited and multiple protocols are required to share VCC instances.

PPP adopts a multi-protocol encapsulation format similar to atm llc multiplexing. It uses a 2-byte protocol identification domain as its Encapsulation Header. For most of them, the multi-protocol encapsulation capabilities of PPP and ATM are the same.

6.8 Fault Tolerance

ATM can recover from the error link and switch to the backup route. It uses the dynamic routing protocol PNNI signaling, PNNI provides the re-routing capability for establishing connections during initialization and the re-routing capability for established connections when a network failure occurs. It provides error recovery at the ATM layer.

PPP does not provide any fault tolerance capability because it runs on a single link, but the SDH in its lower layer has the automatic protection switching function. When the main ring fails, it can be switched to the standby ring. This capability is also available when the lower layer of an ATM uses SDH.

7 Application Prospects

In essence, IP over SDH is used, and routers are connected through a fast point-to-point link. IP over SDH is used, and routers are connected through a network with multiple collection links, these links carry multiplexing connections, and each connection is related to flexible bandwidth and negotiated service quality. The main difference between the two technologies is the comparison of speed and flexibility, in a specific situation, a factor will play a decisive role.

7.1ISP Backbone

The typical requirement of the ISP Backbone Network is to inform and interconnect the backbone routers to maximize the packet throughput. For this reason, the ISP and its users prefer to use IP over SDH to interconnect the routers. However, due to the lack of effective bandwidth management, service quality, and flexible network engineering for IP over SDH, this enthusiasm is gradually fading. In addition, the fast packet transmission capability provided by IP over SDH may exceed the processing capability of most routers. However, IP over SDH has many advantages in WAN applications with limited bandwidth and lax QoS requirements.

7.2 enterprise intranet

Cross-regional enterprise networks are facing the same problems as ISP Backbone Networks. IP over SDH has many advantages in terms of cost, but these advantages must be weighed against the costs required for obtaining the appropriate equipment and services. The operator will sell more ATM links than SDH links, because the ATM has flexible bandwidth management functions. In addition, ATM allows multi-protocol applications, including IP applications and non-IP applications, to share bandwidth in a wide area.

7.3 Campus Network

Campus network does not use SDH because there are physical interfaces cheaper than SDH, such as multimode optical fiber, shielded twisted pair STP, and unshielded cat5e twisted pair UTP. Even if SDH is used, the bandwidth is very plentiful, the bandwidth efficiency advantage of IP over SDH is no more flexible than that of IP over ATM.

7.4 operator Network

The operator has basically adopted SDH as its network base. The operator will adopt ATM over ATM to provide flexible bandwidth management to ensure service quality for its paying users. Compared with the direct operation of IP on SDH, it is easier for carriers to provide IP over ATM.

8 conclusion

There are two obvious trends in today's network industry: one is driven by the internet, IP rapidly becomes a network layer technology, and there is a chance to build a packet-based network, the other is that SDH has been widely used by operators and has become the infrastructure of physical networks. routers can be connected within a wide area based on this.
ATM or PPP can be used between IP and SDH. Which transmission technology is currently unclear, but the speed and flexibility will be the deciding factor for choosing ATM or POS technology. In applications where speed is the primary indicator, IP over SDH is more attractive. IP over ATM is the best solution for flexible bandwidth management, QoS, and network engineering.