CHAPTER 33 WAN

Dear teacher,
IBM SYSTEMS NETWORK ARCHITECTURE (SNA)

The most prominent example of multidrop architecture is IBM’s SNA, which is a tree-structured hierarchical architecture, with a mainframe computer at the head. The architecture was first published in 1974 when processing was controlled by an expensive mainframe computer such as an IBM 360 that communicated with remote 3270 terminals over a network composed of voice-grade circuits.The architecture provides error-free communications over error-prone narrow bandwidth analog circuits, while still providing the data integrity needed for corporate transactions such as bank transfers, reservations, and payroll.

SNA PHYSICAL COMPONENTS

The physical components in an SNA network are hosts, communications controllers (also known as front-end processors), cluster controllers, and terminals. The host provides database access, computations, directory services, network management, and program execution. Front-end processors control the network,and cluster controllers manage the terminals and printers. In the original network a mainframe computer runs a control program known as Advanced Communications Function/Virtual Telecommunications Access Method (ACF/VTAM), usually abbreviated VTAM. VTAM maintains a table of every device and circuit in its domain and controls information flow through the network. It establishes sessions and activates and deactivates resources. Communications with these resources are handled through the mini computerfront -end processors, which in turn communicate with cluster controllers and terminals. Figure 33-3 is a diagram of the logical and physical elements in SNA. SNA establishes a logical path between network nodes and routes each
message with addressing information contained in the protocol. The datalink protocol is SDLC, which bears close resemblance to HDLC.

SNA’S LAYERED ARCHITECTURE

 SNA is defined in layers that are roughly analogous to the layers in ISO’s OSI model. SNA was the basis for much of the OSI model, but the layers do not line up exactly and it differs from OSI in several other respects.
 Level 1, Physical, is not part of the SNA architecture. The physical interface for analog voice-grade circuits is ITU V.24 and V.31. The digital interface is X.21.
 Level 2, Datalink Control, is SDLC. Later versions added support for X.25, token ring, Ethernet, frame relay, and FDDI. The SDLC frame, which is identical in structure to the HDLC frame in Figure 4-6, has six octets of overhead. The first octet is a flag to establish the start of the frame. This is followed by a one-octet address and a one-octet control field. Next is a variable length data field followed by a two-octet CRC field and an ending flag, which becomes the starting flag of the next frame. The control field contains the number of packets received to allow SDLC to acknowledge multiple packets simultaneously. SDLC permits up to 128 unacknowledged packets, which enables it to function with satellite circuits. This layer corresponds closely to ISO’s datalink layer and the LAPB protocol used in X.25 networks.
 Level 3, Path Control, establishes data paths through the network. It carries addressing, mapping, and message sequencing information. At the start of a session, the path control layer establishes a virtual route, which is the sequence of nodes forming a path between the endpoints. The circuits between the nodes are formed into transmission groups, which are circuits having identical characteristics such as speed, delay, and error rate. The path control layer is also responsible for address translation. Through this layer, LUs can address other LUs without being concerned with the entire detailed address of the other terminal. This layer is also responsible for flow control, protecting the network’s resources by delaying traffic that would cause congestion. The path control layer also segments and blocks messages. Segmenting is the process of breaking long messages into manageable size. Blocking is the reverse—combining short messages so the network’s resources are not consumed by small messages of uneconomical size.
 Level 4, Transmission Control, is responsible for pacing. At the beginning of a session the LUs exchange information about variables, such as transmission speed and buffer size, that affect their ability to receive information. The pacing function prevents a LU from sending more data than the receiving LU can accept. Through this layer, SNA also provides other functions such as encryption,message sequencing,and flow control
 Level 5, Data Flow Control, conditions messages for transmission by chaining and bracketing. Chaining is the process of grouping messages with one-way transmission requirements, and bracketing is grouping messages for two-way transmission.
 Level 6, Presentation Services, has three primary purposes. Configuration service activates and deactivates internodal links. Network operator service is the interface through which the network operator sends commands and receives responses. The management services function is used in testing and troubleshooting the network.
 Level 7, Transaction Services, is responsible for formatting data between display devices such as printers and CRTs. It performs some functions of the ISO presentation layer, including data compression and compaction. It also synchronizes transmissions. SNA lacks an applications layer as such, but IBM has defined standards that allow for document interchange and display between SNA devices. Document interchange architecture can be thought of as the envelope in which documents travel. DIA standards cover editing, printing, and displaying documents. The document itself is defined by document content architecture, which is analogous to the letter within the envelope. The purpose of the DIA/DCA combination is to make it possible for business machines to transmit documents with formatting commands such as tabs, indents, margins, and other format information intact. Documents containing graphic information are defined by graphic codepoint definition, which defines the placement of graphic symbols on printers and screens.

ADVANCED PEER-TO-PEER NETWORK (APPN)

The original SNA protocol was built on the assumption that devices such as terminals and printers lacked processing capability so all traffic flowed through the host. This condition was reasonable when terminals lacked intelligence, but as user devices evolved from dumb terminals to PCs, direct peer-to-peer communication became necessary. Cluster controllers could support only one SDLC link and could not communicate among themselves, so IBM developed a cluster controller modification known as PU 2.1. This enabled two controllers to be linked across an SDLC or dial-up connection without requiring a path through the frontend
processor. PU 2.1 supports the physical connection, but it does not provide all the logical functions necessary for peer-to-peer communications. To enable device-to-device communications, IBM introduced the LU 6.2 Advanced Program-to-Program Communications (APPC) protocol. LU 6.2 severs the SNA master/slave relationship between devices, permitting communication between peers. Either device can manage the session, establish and terminate communications, and initiate session error recovery procedures without involving an SSCP. APPC permits direct PC-to-mainframe communications, which enables the PC to transfer files without consuming excessive mainframe processing power. The new SNA is called APPN. APPN defines two kinds of nodes: network nodes and end nodes. An end node, as the name suggests, can send and receive traffic, but data is not routed through it. A network node handles through traffic, and acts as the concentration point for end nodes. Devices on the network are named. To communicate with another device, an end node sends a bind message to its network node, which in turn broadcasts a query that passes through the network. The node answering to that name responds to the query with a message that establishes the session and route.

X.25 PROTOCOL

X.25, built on the first three layers of the OSI model, specifies the interface between DTE and a packet-switched network. The physical layer interface is X.21. The link layer uses LAPB to control errors, transfer packets, and to establish the datalink. The network layer establishes logical channels and virtual circuits between the PAD and the network. X.75 protocol prescribes the interface for gateways between packet networks. Packet networks offer both PVCs and SVCs. With a PVC a path between users is provisioned, and all packets take the same route through the network. With an SVC, the session is managed through control packets, which are analogous to signaling in a circuit-switched network. For example, a call setup packet would be used to establish the initial connection to the terminating device, which would return answer packets. The network uses control packets to interrupt calls in progress, disconnect, show acceptance of reversed charges, and other such functions.

VERY SMALL APERTURE TERMINAL (VSAT)

VSAT is a Ku band satellite service that is an excellent medium for a widely dispersed
operation. Its pricing is not distance sensitive, so it is particularly effective in remote locations. Typical applications include LAN/WAN networking, Internet access, and videoconferencing. It is economical in rural areas and small towns where frame relay is not cost effective because of the access circuit cost.Another application for which VSAT is well suited is telemetering from mobile devices such as trucks, ships, and trains. Figure 33-4 shows a typical VSAT arrangement. As discussed in Chapter 19, the VSAT terminal is a small device with a 90 to 120 cm dish antenna. Transponder space is leased from the satellite provider or obtained from the hub provider.

WAN APPLICATION ISSUES

Data network applications can be separated into the following general types:
• Inquiry/response: This is typical of information services where a short inquiry generates a lengthy response from the host. Because the data flow is asymmetric, half-duplex facilities generally offer the greatest throughput on a dedicated private line. On digital facilities, the connection is inherently full duplex. Typical applications are airline reservations, streaming video, and on-line database sessions such as the World Wide Web. The operator keys a few characters into the terminal, and the host computer responds with a lengthy message that might be confirmation of a reservation, a printed ticket, or an information dump.
• Conversational: This mode, typical of terminal-to-terminal communication, is characterized by short messages that are of approximately equal length in both directions. Throughput is improved by using full-duplex operation. Conversational mode is typical of voice and videoconference over IP.
• Bulk data transfer: This is typical of applications such as storage area networks where large files are passed, often at high speed, in only one direction. This method is often used when a local processor collects information during the day and makes daily updates of a master file such as an inventory on the host.
• Remote job entry: In RJE remote terminals send information to a host. The bulk of the transmission is from the terminal with a short acknowledgment from the host. Half-duplex circuits may be the most effective form of transmission because the bulk of the information flows from remote to host. Dedicated facilities, either leased lines or frame relay, are almost invariably needed for this kind of application. Many remote terminals, each of which is used only occasionally, may share a higher speed line to the host.