HP OpenVMS Systems Documentation

CSMA/CD drivers support the following features:

• Ethernet and IEEE 802.3 packet format
• Physical layer identified as type 10Base5 (10-Mbps baseband (ThickWire)), 10Base2 (ThinWire), and 10BaseT (twisted pair)
• Fast Ethernet and IEEE 802.3u packet format
• Fast Ethernet physical layer identified as type 100BaseT
• Gigabit Ethernet and IEEE 802.3z packet format
• Gigabit Ethernet's physical layer is identified as type 1000BaseT for unshielded twisted pair (UTP), category 5, 1000Base-SX for multimode fiber-optic cables.

FDDI drivers support the following features:

• FDDI packet format
• Transmission and reception of frame control (FC) priority

Token Ring drivers support the following features:

• IEEE 802.5 packet format

ATM drivers over ELAN support the following features:

• Ethernet and IEEE/802.3 packet format
• UNI Version 3.0 or 3.1 signaling protocol
• LAN emulation (LANE) Version 1.0
• Maximum frame sizes of 1516, 4544, and 9234 bytes

All LAN drivers support the following features:

• 802.2 packet format
• IEEE 802.2 Class I service including the unnumbered information (UI), exchange identification commands and responses (XID) and TEST commands, and the XID and TEST responses
• IEEE 802.2 Class II service may be specified where the functions are provided by the user application
• 6-byte destination and source address fields
  The IEEE 802.3 Standard states that the size of the destination and source addresses be 2 or 6 bytes, as decided by the manufacturer. The OpenVMS LAN drivers and controllers only support 6-byte address fields.

Contrary to the IEEE 802.2 Standard, the Global DSAP (FF) must beenabled as a Group SAP to receive messages with the Global DSAP in thedestination SAP field.

FDDI conforms to the ANSI Standards defined in the following documents:

• ANSI X3.139-1987 FDDI Media Access Control (MAC)
• ANSI X3.148-1988 FDDI Physical Layer Protocol (PHY)
• ANSI X3.166-1990 FDDI Physical Layer Medium Dependent (PMD)

System software and user applications communicate with other systemsthrough the LAN controllers using the QIO interface defined by theOpenVMS LAN driver software. The description of this interface isdescribed later in this chapter. The LAN driver software allowscommunication with these differing technologies in a consistent manner.

The LAN controllers implement the Ethernet, FDDI, Token Ring, and ATMspecifications. The Ethernet specification is described in TheEthernet-Data Link Layer and Physical Specification. The FDDIspecifications are available from ANSI. The Token Ring specificationsare available from IEEE. The ATM LAN Emulation specifications areavailable from the ATM Forum. The Classical IP over ATM specification(RFC 1577) is available from the Internet Engineering Task Force (IETF).

Ethernet includes Fast Ethernet (802.3u) and Gigabit Ethernet (802.3z).See Section 9.4 and Section 9.5 for more information.

Ethernet, FDDI, Token Ring, and ATM networks can be configured to forma single extended LAN using FDDI-Ethernet bridges, FDDI and Ethernetswitches, Token Ring bridges and routers, and ATM switches. This allowsapplications running on a system connected by a LAN controller of onetechnology to communicate with applications running on another systemconnected by a different type of LAN controller.

9.4 Fast Ethernet LANs (Alpha Only)

Fast Ethernet (802.3u) is an extension of the IEEE 802.3 standard. Ittypically runs over twisted-pair wiring (100BaseT). It increases thedata transmission rate from 10 to 100Mbps and decreases the maximumlength of a network segment. Fast Ethernet controllers allow either 10or 100Mbps operation for compatibility with existing 10Mbps controllerson the same network segment. The DE500-FA Fast Ethernet controllerprovides a fiber-optic connection (100BaseFX).

Table 9-3 shows the types of cabling used for Fast Ethernet.

The OpenVMS operating system supports the following Fast Ethernetadapters on Alpha PCI-based systems:

DE500-XA
DE500-AA
DE500-BA
DE500-FA

Gigabit Ethernet (802.3z) is an extension of the IEEE 802.3 standard.It runs over fiber-optic cabling and twisted-pair wiring. It increasesthe data transmission rate to 1000Mbps. The frame formats are identicalto Ethernet and Fast Ethernet which allows good interoperability acrossthese technologies. Gigabit Ethernet is suitable as a high-speedbackbone interconnect but may be used to connect high-performanceworkstations or systems that need the increased bandwidth. Twisted-pairGigabit controllers allow either 10, 100, or 1000 Mbps operation forcompatibility with existing 10 or 100 Mbps controllers on the samenetwork seqment.

Table 9-4 shows the types of cabling used for Gigabit Ethernet.

OpenVMS supports the DEGPA Gigabit Ethernet LAN controller on AlphaPCI-based systems.

9.6 LAN ATM Network Support (Alpha Only)

Asynchronous Transfer Mode (ATM) is a cell-oriented switchingtechnology that uses fixed length packets to carry different types ofdata.

The ATM protocol communicates by first establishing endpoints betweentwo computers with a virtual circuit (VC) through one or more ATMswitches. ATM then provides a physical path for data flow between theendpoints by either a permanent virtual circuit (PVC), or a switchedvirtual circuit (SVC).

Permanent Virtual Circuits (PVCs)

Permanent Virtual Circuits are set up and torn down by priorarrangement. They are established manually by a user before the sendingof any data between endpoints on a network. Some PVCs are defineddirectly on the switch; others are predefined for use in managingswitched virtual circuits (SVCs).

Switched Virtual Circuits (SVC)

Switched virtual circuits require no operator interaction to create andmanage connections between endpoints. Software sets up and tears downconnections dynamically as they are needed through the request of anendpoint.

OpenVMS has deployed ATM networks based on the ATM LANE standards andClassical IP over ATM (RFC 1577). The following ATM adapters on Alphasystems are supported by OpenVMS with the ATM LANE standards:

DGLTA
DGLPB
DGLPA
DAPBA
DAPCA

The following ATM adapters on Alpha systems are supported by OpenVMSwith Classical IP over ATM (RFC 1577):

DGLTA
DBLPB
DGLPA

LAN emulation over an ATM network allows existing applications to runessentially unchanged while also allowing the applications to run oncomputers directly connected to the ATM network. The LAN emulationhides the underlying ATM network at the media access control (MAC)layer, which provides device driver interfaces.

Table 9-5 shows the four components that make up a LAN emulationover ATM network. Of the four components, OpenVMS supports only the LANemulation client (LEC).

The LEC exists on all ATM-attached computers that participate in theLAN emulation configuration. LEC provides the ATM MAC-layerconnectionless function that is transparent to the LAN-typeapplications. The LEC, LES, and BUS can exist on one ATM-attachedcomputer or on separate computers. The server functions usually resideinside an ATM switch, but can be implemented on client systems.

9.6.2 LAN Emulation Topology

Figure 9-1 shows the topology of a typical emulated LAN over ATM.

Figure 9-1 Emulated LAN Topology

Classical IP (CLIP) implements a data-link level device that has thesame semantics as an Ethernet interface (802.3). This interface is usedby a TCP/IP protocol to transmit 802.3 (IEEE Ethernet) frames over anATM network. The model that OpenVMS Alpha follows for exchanging IPdatagrams over ATM is based on RFC 1577 (Classical IP over ATM).

For information on using LANCP commands to manage Classical IP, referto the OpenVMS System Management Utilities Reference Manual: A--L.

9.7 Supporting and Configuring LAN Emulation over ATM (Alpha Only)

OpenVMS provides LAN Emulation Client (LEC) support over ATM. The LANEmulation Client software supports IEEE/802.3 Emulated LANs, and UNI3.0 or UNI 3.1 and the following maximum frame size (in bytes): 1516,4544, and 9234.

The DAPBA (155 Mb/s) and the DAPCA (622 Mb/s) are ATM adapters forPCI-bus systems that are supported by SYS$HWDRIVER4.EXE. The followingrequirement applies to the DAPBA and DAPCA adapters:

Both adapters require a great deal of non-paged pool, and therefore,care should be taken when configuring them. For each DAPBA, Compaqrecommends increasing the SYSGEN parameter NPAGEVIR by 3000000. Foreach DAPCA, Compaq recommends increasing NPAGEVIR by 6000000. To dothis, add the ADD_NPAGEVIR parameter to MODPARAMS.DAT and then runAUTOGEN. For example, add the following command to MODPARAMS.DAT on asystem with two DAPBAs and one DAPCA:

                     ADD_NPAGEVIR = 12000000

The following restrictions apply to the DAPBA and DAPCA adapters:

The adapter cannot be located on a PCI bus that is located behind a PCI-to-PCI bridge. Systems that have this configuration are the following:

Digital Personal AlphaWorkstation 600 (MIATA GL)
AlphaStation 1000A (Noritake)
COMPAQ Professional Workstation XP1000 (MONET)
Alphaserver 2000 and 2100 (SABLE)

SYS$LAN_ATM4.EXE provides OpenVMS ATM infrastructure for the DAPBA andDAPCA adapters. SYS$ELDRIVER4.EXE provides the Emulated LAN support forthe DAPBA and DAPCA adapters.

The DGLPB (155 M/bs) is an ATM device for PCI-bus systems that issupported by SYS$HWDRIVER.EXE.

The DGLPA (155 M/bs) is an ATM device for PCI-bus systems that issupported by SY$ATMWORKS351.EXE.

The DGLTA (155 M/bs) is an ATM device for Turbochannel systems with theexception of the DEC 3000-300 that is supported by SYS$HCDRIVER.EXE.

SYS$LAN_ATM.EXE provides the OpenVMS ATM infrastructure for the DGLPB,DGLPA, and DGLTA adapters. SYS$ELDRIVER.EXE provides the Emulated LANsupport for the DGLPB, DGLPA, and DGLTA adapters.

The Emulated LAN driver provides the means for communicating over theLAN ATM. The device type for the Emulated LAN device is DT$_EL_ELAN.

The device name for the Emulated LAN is:

        ELcu    where c is the controller and u is the unit number (for example, ELA0).

The ATM software is set to autosense the UNI version by default.Setting bit 3 of the system parameter, LAN_FLAGS, to 1 enables UNI 3.0over all ATM adapters. Setting bit 4 of the system parameter,LAN_FLAGS, to 1 enables UNI 3.1 over all ATM adapters.

9.7.2 Enabling SONET/SDH

The ATM drivers have the capability of operating with eithersynchronous optical network (SONET) or synchronous digital hierarchy(SDH) framing. Setting bit 0 of the system parameter, LAN_FLAGS, to 1enables SDH framing. Setting bit 0 of the system parameter, LAN_FLAGS,to 0 enables SONET framing (default). For this to take affect, thesystem parameter must be specified correctly before the ATM adapterdriver is loaded.

9.7.3 Booting

OpenVMS Alpha does not support ATM adapters as boot devices.

9.7.4 Configuring an Emulated LAN (ELAN)

The LANCP utility sets up an Emulated LAN (ELAN). If the ELAN isdefined in the permanent database, these settings take affect at boottime. To define the commands in the permanent database for specificadapters, you invoke the DEFINE DEVICE commands. Once these commandsdefine the adapters in the permanent database, the ELAN can be startedduring system startup.

You can also invoke the LANCP SET commands to start up an ELAN afterthe system is booted.

The following example shows the DEFINE DEVICE commands that define theadapter in the permanent database.

$ mcr lancpLANCP> define device ela0/elan=createLANCP> define device ela0/elan=(parent=hwa0,type=csmacd,size=1516)LANCP> define device ela0/elan=(descr="An ATM ELAN")LANCP> define device ela0/elan=enable=startupLANCP> list dev ela0/paramDevice Characteristics, Permanent Database, for ELA0:                  Value  Characteristic                  -----  --------------                   HWA0  Parent ATM device          "An ATM ELAN"  Emulated LAN description                   1516  Emulated LAN packet size                CSMA/CD  Emulated LAN type                    Yes  Emulated LAN enabled for startupLANCP> exit$

The following example shows the SET DEVICE commands required forsetting up an ELAN with the desired parameters. Note that some of thecommands generate a console message.

$ mcr lancpLANCP> set dev ela0/elan=create%%%%%%%%%%%  OPCOM  26-MAR-2001 16:57:12.89  %%%%%%%%%%%Message from user SYSTEM on ALPHA1LANACP LAN ServicesFound LAN device ELA0, hardware address 00-00-00-00-00-00LANCP> set dev ela0/elan=(parent=hwa0,type=csmacd,size=1516)LANCP> set dev ela0/elan=(descr="An ATM ELAN")LANCP> set dev ela0/elan=enable=startup%ELDRIVER, LAN Emulation event at 26-MAR-1996 16:57:28.78%ELDRIVER, LAN Emulation startup: Emulated LAN 1 on device ELA0LANCP> sho dev ela/char   Device Characteristics ELA0:                  Value  Characteristic                  -----  --------------                 Normal  Controller mode               External  Internal loopback mode                CSMA/CD  Communication medium                     16  Minimum receive buffers                     32  Maximum receive buffers                     No  Full duplex enable                     No  Full duplex operational            Unspecified  Line media                     10  Line speed (megabits/second)                CSMA/CD  Communication medium                 "HWA0"  Parent ATM Device          "An ATM ELAN"  Emulated LAN Description   3999990000000008002B  LAN Emulation Server ATM Address   A57E80AA000302FF1300                Enabled  Emulated LAN StateLANCP> exit$

For information about using LANCP and system manager commands withqualifiers for LAN emulation over ATM networks, refer to theOpenVMS System Management Utilities Reference Manual: A--L, and OpenVMS System Manager's Manual.

9.8 Ports and LAN Configuration

A port in a LAN configuration consists of a protocoltype, a service access point (SAP) or protocol identifier, and acontroller. There are as many ports on a LAN controller as there areprotocol types, SAPs, and protocol identifiers. Each port isindependent of other ports running on the same LAN controller.

Application programs use either the LAN driver's QIO interface or VCI interface to perform I/O operations to and from other nodes on the LAN. This chapter describes the QIO interface. Figure 9-2 shows the relationship of most Ethernet controllers to the processor and to the user application program.

Figure 9-2 Typical Ethernet Configuration

The following sequence initializes and starts a port on a LAN devicedriver:

1. Use the Assign I/O Channel ($ASSIGN) system service to assign I/O channels to one or more of the LAN device names and devices specified in Table 9-1 and Table 9-2. $ASSIGN creates a new unit control block (UCB), to which the channel for the port is assigned.
2. Start up the port with the set mode function and startup function modifier (see Section 9.16.3.1). You must supply the required P2 buffer parameters listed in Table 9-16.
3. Perform read, write, and sense mode operations as needed.
4. Shut down the port with the set mode function and shutdown function modifier (see Section 9.16.3.4).
5. Use the Deassign I/O Channel ($DASSGN) system service to deassign the I/O channel.

The sample programs described in Section 9.18.2 illustrate how to perform these procedures for Ethernet and IEEE 802 ports.

9.9 Ethernet Addresses

The LAN is a medium for creating a network; it is not a network byitself. The LAN controller and the local system constitute a node.Nodes on the LAN are identified by unique Ethernet addresses. A messagecan be sent to one, several, or all nodes on the LAN simultaneously,depending on the Ethernet address used. You do not have to specify theEthernet address of your own node to communicate with other nodes onthe same Ethernet. However, you do need to know the Ethernet address ofthe node with which you want to communicate.

9.9.1 Format of Ethernet Addresses

An Ethernet address is 48 bits in length. Ethernet addresses arerepresented by the Ethernet standard as six pairs of hexadecimal digits(six bytes), separated by hyphens (for example, AA-01-23-45-67-FF). Thebytes are displayed from left to right in the order in which they aretransmitted; bits within each byte are transmitted from right to left.In this example, byte AA is transmitted first; byte FF is transmittedlast. (See the description of NMA$C_PCLI_PHA in Table 9-16,Section 9.16.3.1, for the internal representation of addresses.)

Upon application, IEEE assigns a block of addresses to a producer ofLAN nodes. Thus, every manufacturer has a unique set of addresses touse. Normally, one address out of the assigned block of physicaladdresses is permanently associated with each controller (usually inread-only memory). This address is known as the hardware address of thecontroller. Each individual controller has a unique hardware address.