HP OpenVMS Systems Documentation

Content starts here

DECnet-Plus for OpenVMS
Introduction and User's Guide


Previous Contents Index

3.7.2.1 Compliance with OSI Packet Formats and Addressing

An OSI administrative domain consists of a combination of end systems, intermediate systems, and subnetworks. You can also subdivide the administrative domain into routing domains, in which all systems and subnetworks use the same routing protocols.

The Network layer complies with the ISO standards for network packet formats and addressing. Data messages are forwarded through the network in self-contained packets that include the source and destination addresses.

DECnet-Plus for OpenVMS systems support the packet format that is specified in ISO standard 8473, and can exchange data with other vendors' systems that also conform to the ISO packet format standard.

The Network layer accepts messages from the Transport layer and encloses them in packets called network protocol data units (NPDUs). The NPDU includes the Network layer header that contains the source and destination addresses for the data. The Network layer address for a system is called the network service access point (NSAP). The NSAP is located at the boundary between the Network layer and the Transport layer, where communication between the layers takes place. For complete information about NSAPs, see the DECnet-Plus Planning Guide.

The Network layer uses the destination NSAP address to forward the NPDUs to the destination system. The Network layer also provides compatibility with Phase IV.

3.8 Data Link Layer

The Data Link layer provides a communications path between directly connected systems in a network. It controls the movement of information between systems, including the transmission and receipt of data. In providing delivery of data to the adjacent node, the Data Link layer performs some or all of the following functions: establishment of the link (initializing and conditioning the line), error detection and recovery, data flow control, data framing control, and packet sequence control.

DECnet-Plus for OpenVMS uses synchronous, asynchronous (VAX only), Ethernet, and FDDI communications controllers to interface with other network nodes.

LAN connectivity is provided by the CSMA-CD and FDDI controllers and drivers supporting ISO 8802-2 logical link control (LLC) type 1 connectionless service and ISO 8802-3 LLC type 2. DECnet-Plus also supports Ethernet V2 packets on CSMA-CD devices.

Use of FDDI packets larger than 1500 bytes requires a Phase V router on the FDDI LAN. As with cluster alias support, the Phase V router may be configured to run the Phase IV distance vector routing protocol or the Phase V Link State Routing Protocol.

WAN connectivity is provided by WAN device drivers supporting host-based synchronous communications options for wide area networking.

All the data link devices support DDCMP, HDLC/LAPB and SDLC protocols. BISYNC and GENBYTE (VAX only) protocols are also supported on some options. WAN device drivers are required by X.25 to establish host-based wide area connections.

Synchronous controllers use DDCMP or HDLC, either when directly connected or when connected via modems, to provide full- or half-duplex communications over point-to-point lines. Synchronous DDCMP multipoint tributary connections are also supported. Asynchronous controllers (on VAX systems) use DDCMP, either when directly connected or when connected via modems, to provide only full-duplex communications over point-to-point lines. Error correcting and data suppression modems are not supported.

Asynchronous lines (on VAX systems) are supported only to other systems running DECnet-Plus for OpenVMS VAX, DECnet-VAX, DECnet-RSX, and DECnet-DOS.

DDCMP operation (on VAX systems) is not supported in cases where an asynchronous physical communications line is emulated by lower-level protocols or communications subsystems. Examples of this include X.29 virtual terminals, asynchronous connections as emulated by terminal servers, and connections via data switches.

DECnet-Plus for OpenVMS allows up to four circuits to be defined and operational on an end system. This capability allows a single end system to be connected to up to four separate LANs or WANs. Note that all circuits must be equal in capacity and connectivity.

3.8.1 Support for CSMA-CD Protocol on a LAN

The Data Link layer allows the transmission of data over a local area network cable by means of the CSMA-CD (Carrier Sense Multiple Access with Collision Detection) protocol. CSMA-CD ensures equal access to multiple systems connected to the LAN. DECnet-Plus for OpenVMS supports both the existing Ethernet protocol and the protocol that complies with ISO 8802 (also known as IEEE 802).

The Ethernet and ISO 8802 protocols are compatible, with only slight differences in packet format. A DECnet-Plus system transmits in both ISO and Ethernet formats, and listens for frames in ISO and Ethernet formats. If a Phase IV system is connected to the LAN, the DECnet-Plus system also transmits in Ethernet format.

CSMA-CD LANs are similar to LANs defined by ISO 8802-3. LANs are privately owned, reliable, high-speed networks that connect information-processing systems in a limited geographic area, such as an office, a building, or a complex of buildings. It is a best-effort delivery system.

CSMA-CD allows multiple stations to access the broadcast channel at will, avoids contention by means of carrier sense and deference, and resolves contention by means of collision detection and retransmission. Each station awaits an idle channel before transmitting and can detect overlapping transmissions by other stations.

CSMA-CD stations provide for multiaccess connection between a number of systems on the same broadcast circuit. This kind of routing circuit is a path to many systems. Each system on one CSMA-CD station is considered adjacent to every other system on that station and equally accessible.

The media is passive coaxial cable or shielded twisted-pair cable that uses Manchester-encoded, digital baseband signaling, with interconnections containing all active components so that no switching logic or central computer is needed to establish or control communications.

At the Data Link layer, network control for the LAN is multiaccess, fairly distributed to all systems. The frame length allocation is from 64 to 1518 bytes (including an 18-byte envelope).

A system on an ISO 8802-3 (CSMA-CD) LAN is connected to the CSMA-CD station by:

  • A LAN communications controller
  • A transceiver
  • A transceiver cable

When manufactured, LAN controllers are given a 48-bit hardware address.

A particular ISO 8802-3 system is identified by the hardware address of its station (line); this hardware address is stored in read-only memory in the LAN controller. When DECnet-Plus starts a CSMA-CD station, it constructs a physical address for the system. When you shut off machine power or change the state of the CSMA-CD station to off, the LAN controller resets the physical address to the original hardware address.

DECnet-Plus has no restrictions on the number of end systems on a LAN. In addition, you do not need an intermediate system on a CSMA-CD LAN.

  1. When no intermediate system is on the LAN:
    • An end system trying to communicate with another end system sends the data by using a specific multicast address.
    • Only a system whose link service access point (LSAP) matches the destination DSAP in the multicast packet will respond. The destination system responds with an end system hello, sent directly to the originating system, informing the source system of its address. (For definitions of LSAP and DSAP, see Section 3.3.5.2.)
    • The source system and the destination system speak directly because each knows the LAN address of the other.
  2. When one or more intermediate systems is on the LAN:
    • The first message from the source is sent to a randomly selected intermediate system.
    • The intermediate system detects that both the source and destination systems are on the same LAN.
    • The intermediate system forwards that first message to the destination and also sends a redirect message to the source system.
      In this way, end systems learn the LAN addresses of other end systems so they can speak directly.

ISO 8802-3 (CSMA-CD) LANs support a bus topology, a single communications medium to which all the systems are attached as equals. The single network cable replaces the numerous interconnecting cables usually required in traditional WANs. This type of network is also called a broadcast LAN. The maximum possible distance between systems on the LAN is 2.8 kilometers (1.74 miles).

3.8.1.1 Extended LANs

You can connect segments of coaxial cable to extend LANs beyond the 500-meter (1640 feet) limit of a single segment. Extended LANs create larger networks in terms of distance and also in terms of the number of connections that can be made. (A 500-meter segment supports 100 physical connections.) Repeaters and bridges join cable segments:

  • A repeater simply extends the LAN by retransmitting all network traffic that originates on one segment onto the attached segment, enabling connected segments to function as one cable.
  • A bridge, in addition to extending the LAN, filters network traffic between segments.
    If data packets are addressed to a destination system on the attached segment, the bridge transmits the packet to the segment. If data packets are addressed to a local destination system, the packets are confined to the segment to which the source system is attached. The primary benefit of bridges is to reduce network traffic.

3.8.1.2 Intermediate System

If the end systems communicate only with each other, you do not need an intermediate system on a LAN, but can use host-based routing. Optionally, you can use one to limit multicast traffic. To route messages off the LAN over other routing circuits, you must configure an intermediate system such as DDCMP circuits.

If a LAN is operating with more than one area and with one or more level 1 intermediate systems, you need a level 2 intermediate system to transport messages between areas.

3.8.1.3 Multicircuit End Systems

DECnet-Plus for OpenVMS allows multiple circuits to be active and usable simultaneously on an end system. For example, you can connect an end system to two LAN cables. Both routing circuits are used and traffic is split between the circuits, but no routing occurs over these circuits. A DECnet-Plus for OpenVMS end system can support a maximum of three circuits. This feature provides for redundancy and increased data throughput without requiring an intermediate system.

3.8.1.4 Areas and Multihomed Systems

You can partition a large DECnet-Plus routing domain into subdomains called areas. For Phase IV, an area is a group of network nodes that can run independently, with all nodes in the group having the same area address. DECnet-Plus areas are similar to Phase IV areas except for the following new features:

  • Multihomed systems, which have more than one area address
  • Single-area LANs

A node can be in only one area in a network. DECnet-Plus systems, however, can have more than one area address. Such a system is a multihomed system. You can assign up to three area addresses to a DECnet-Plus system.

A DECnet-Plus area is a set of systems that all share the same area address (or addresses). An area (and the systems within the area) can have more than one area address. For example, if an area in a DECnet-Plus network is connected to an X.25 public network, a system in that area of the network would have two addresses: one for the DECnet-Plus network and one for the X.25 network. A system cannot have addresses on two networks that are not connected.

If you connect two areas to a DECnet-Plus LAN, the level 2 intermediate systems automatically combine themselves into one area with two area addresses. Phase IV LANs can be divided into several areas. Mixed Phase IV and DECnet-Plus LANs can also include several areas.

3.8.2 Support for the HDLC Protocol

High-level Data Link Control (HDLC) protocol data links are ISO, synchronous, point-to-point links that are basically the same in function as existing DDCMP synchronous links. However, HDLC is a bit-oriented protocol, whereas DDCMP is a byte-oriented protocol.

HDLC operates over synchronous, switched, or nonswitched communications links. HDLC supports a broad range of existing subsets, including the subset used in X.25 networks.

HDLC operates in either of two modes:

  • Balanced mode --- Operational mode used over full-duplex links
  • Normal mode --- Operational mode used over half-duplex links

HDLC links use UI frames and XID frames. A UI frame is an unnumbered, information frame that carries data not subject to flow control or error recovery. An XID frame exchanges operational parameters between participating stations.

3.8.2.1 LAPB Support

DECnet-Plus for OpenVMS also supports a modified form of HDLC called link access protocol balanced (LAPB). LAPB is the CCITT-approved link level protocol for X.25 connections. LAPB defines the procedure for link control in which the DTE/DCE interface is defined as operating in two-way asynchronous balanced mode (ABM). LAPB is for the reliable transfer of a packet from a host to an X.25 packet switch, which then forwards the packet on to its destination.

3.8.3 Support for the DDCMP Protocol

DDCMP is designed to provide an error-free communications path between adjacent systems. It operates over serial lines, delimits frames by a special character, and includes checksums at the link level.

DECnet-Plus for OpenVMS continues to support proprietary DDCMP data links, which include these types of connections:

  • Synchronous point-to-point
  • Synchronous tributary multipoint
  • Asynchronous point-to-point
  • Asynchronous static (permanent)
  • Asynchronous dynamic (switched temporary)

DDCMP provides a low-level communications path between systems. The protocol detects any bit errors that are introduced by the communications channel and requests retransmission of the block. The DDCMP module provides for framing, link management, and message exchange (data transfer). Framing involves synchronization of bytes and messages.

DDCMP pipelining permits several packets to be sent before an acknowledgment is received. Piggybacking permits an acknowledgment to be transmitted on a data packet.

The DDCMP protocol moves information blocks over an unreliable communication channel and guarantees delivery of routing messages. Individual systems on DDCMP routing circuits are addressed directly because no multicast or broadcast addressing capability is available.

The Data Link layer supports point-to-point, DDCMP links, either synchronous or asynchronous. The two types of asynchronous links are static (permanent) and dynamic (switched temporary).

3.8.3.1 Synchronous DDCMP

Synchronous links provide the medium to high-speed point-to-point communication. The synchronous DDCMP Protocol can run in full- or half-duplex mode. This allows DDCMP the flexibility of being used for local synchronous communications or for remote synchronous communications over a telephone line using a modem.

DDCMP is implemented in the driver software (WANDD) for the synchronous communications port.

3.8.3.2 Asynchronous DDCMP

Asynchronous links provide a low-speed, low-cost media for point-to-point communication. Asynchronous DDCMP can run over any directly connected station that the DECnet-Plus for OpenVMS system supports. Asynchronous DDCMP provides for a full-duplex connection. You can use it for remote asynchronous communications over a telephone line using a modem. Asynchronous connections are not supported for maintenance operations or for controller loopback testing.

Asynchronous DDCMP does not need to be predefined for dynamic connections. It is established automatically when a dynamic asynchronous DDCMP link is made.

DDCMP asynchronous links can be static or dynamic.

  • Static connection --- The asynchronous station is permanently configured as a communications link.
    A static asynchronous DDCMP connection is a permanent DECnet-Plus connection between two systems physically connected by stations. You convert them to static asynchronous DDCMP stations by issuing NCL commands to set the stations to support the DDCMP protocol. The user at each system then turns the appropriate routing circuits and stations on for DECnet-Plus use. After the communications link is established, it remains available until a user turns off the routing circuit and station and clears the entries from the appropriate NCL script file.
    Static asynchronous DDCMP configurations require the asynchronous DDCMP driver to be connected. The asynchronous DDCMP protocol can run in full-duplex operation on local asynchronous communication links.
    You can configure a dial-up line as either a static or dynamic asynchronous station, but the dynamic connection may be more secure and convenient to use.
  • Dynamic connection --- A station connected to a terminal port is switched to an asynchronous communications station for the duration of a call.
    A dynamic asynchronous connection is a temporary connection between two systems, generally over a telephone line using modems. The stations at both ends of the connection can be switched to asynchronous DDCMP communications stations and then switched back to terminal lines.
    You can use dynamic asynchronous connections to establish a DECnet-Plus link to another system for a limited time or to create links to different systems at different times.

3.8.3.3 Converting DDCMP Links to HDLC Links

DECnet-Plus supports HDLC as an alternative to DDCMP to:

  • Align DECnet-Plus more closely with international standards
  • Support new industry-standard communications controller chips with higher-level capability

You can convert existing DDCMP point-to-point synchronous lines to HDLC lines. However, DECnet-Plus also supports DDCMP for compatibility and to provide these capabilities not available with HDLC:

  • Continuing support for existing controllers that cannot use HDLC
  • Use over asynchronous lines

3.9 Physical Layer

The Physical layer is responsible for the transmission and receipt of data on the physical media that connects systems. It transparently moves data between the system and the communications path signaling equipment. The Physical layer can include part of the device driver for a communications device and for communications hardware: interface devices, modems, and communication lines.

3.9.1 CSMA-CD LAN Interface

The Physical layer supports the CSMA-CD interface, which complies with ISO standard 8802-3 and the IEEE 802.3 standard. The Physical layer also allows CSMA-CD LAN connections based on the DECnet Phase IV Ethernet interface.

3.9.2 Modem Connect Module

The DECnet-Plus Modem Connect module defines the operation of synchronous and asynchronous devices. It provides for network management of stations (physical lines) that conform to industry standards for modem connection. You can establish and monitor the following types of links:

  • Synchronous and asynchronous connections
  • Point-to-point lines
  • Tributary multipoint lines
  • Leased lines

The Modem Connect module supports several industry standards for physical interfaces:

  • Electronic Industries Association (EIA) RS-232-C, RS-422, and RS-423
  • CCITT V.24
  • CCITT V.25 and V.25bis for auto call and auto answer
  • CCITT V.35

The Modem Connect module does not contain driver-specific code. It contains all the common routines related to network management for all synchronous and asynchronous drivers. The functions controlled through the Modem Connect module include:

  • Data transfer services
  • Network management of the port and line entities
  • Line profile identification which defines any nonstandard operation of the line

3.10 Network Management

DECnet-Plus for OpenVMS network management capabilities include:

  • The director-entity model.
  • A command line management interface, the Network Control Language (NCL), which replaces the Phase IV Network Control Program (NCP).
  • Remote management of Phase IV nodes using the NCP Emulator.
  • DNA Common Management Information Protocol (DNA CMIP).
  • DECnet-Plus for OpenVMS initialization scripts support.
  • Maintenance operations: downline load, upline dump, remote console connection, and loopback testing support.
  • An enhanced event logging (EVD).
  • Common Trace Facility (CTF) for troubleshooting.
  • Network management graphical user interface (NET$MGMT), an enhanced Motif-based windows interface.

3.10.1 Network Management Components

The structure of DECnet-Plus network management defines formal relationships between the management software at the various layers and the directors that communicate with the layers on behalf of network managers.

The two major components of network management are directors and entities. Directors are interfaces for managing the entities. They are typically used by the network manager and usually involve a command language. NCL management entities, which are the manageable components that make up the network, relate to other entities on the same system.

The entity model defines the structure of the entities that constitute a distributed system and the management functions they provide.

  • node entity --- Top-level entity in the entity hierarchy; identified by a globally unique node name, which represents an individual computer system.
  • Module --- The next level of entity on a node; a group of network functions that provide a service. Each module includes the entities that make up the module. An example of a module is a protocol, for example HDLC, and the entities used to manage that protocol. Each entity defines the management functions, characteristics, status, and attributes that are pertinent for its operation.
    There is only one occurrence, or instance, of a module entity in a node. For this reason, modules can be uniquely identified within a node by their class name.
  • Entities --- The next level down in the hierarchy; defined with NCL commands to allow management of some part of a module's functions. The LAPB module, for example, maintains lapb link entities for each communications link over which the protocol operates; lapb link is the full class name of the entity. Each instance of the lapb link entity requires further identification to allow it to be distinguished from the others.

For further information on NCL commands, refer to the DECnet-Plus Network Control Language Reference guide.


Previous Next Contents Index