HP e3000/iX Network Planning and Configuration Guide > Chapter 2 Networking Concepts

Address Resolution

	MPE documents
	Complete PDF
	Table of Contents
	Index

Address resolution in NS networks refers to the mapping of node names to IP addresses and the mapping of IP addresses to lower level addresses (such as an X.25 address or a station address). Several address resolution methods are available for you to use individually or in combination with each other. You can configure these methods according to the needs of your network.

The available address resolution methods are:

Domain name services.
Network directory.
Probe (and probe proxy) (LAN, 100VG-AnyLAN, and 100Base-T only).
Address resolution protocol (ARP) (LAN, Token Ring, FDDI, 100VG-AnyLAN, and 100Base-T only).

Domain Name Services

The domain name services are a mechanism for resolving node names to IP addresses. They conform to an open networking standard and will facilitate communications between HP e3000 systems as well as with non-HP e3000 nodes.

To use the domain name services, you must assign a name, in ARPANET standard format, to each system on the network or internetwork. You configure this name on the NS Configuration screen (see configuration chapters for details).

You will also need to create a set of ASCII files on each system which contain the addressing information the system will need. Instructions for creating these files are in Chapter 12 "Configuring Domain Name Files"

Once you have configured the domain name services, the network will be able to access the node using its domain name and the domain name service routines will resolve the domain name to the node's IP address.

NOTE: Domain name services provide name to IP address resolution only. If a lower level address is required for network communication (for example, an X.25 address) you will need to configure the network directory as well.

Network Directory

The network directory is a set of files that contain information used by the node to communicate with other nodes in the internetwork.

You use NMMGR to perform the following network directory functions:

Add, modify, and delete entries in the directory.
Review and inspect directory information.
Merge a remote directory with a directory on the local node.
Automatically update directories on a group of remote nodes by using a background stream job controlled from a central administrative node.

See Chapter 11 "Configuring the Network Directory" for more information on configuring the network directory through NMMGR. More information on merging directories and on central administrative nodes is included in this chapter.

When a Network Directory is Required

A network directory must be configured in the following circumstances:

nodes running on X.25
nodes not using domain name services
nodes on a LAN network that do not support the HP-PROBE protocol

The network directory of a node in a Point-to-Point network must contain the IP addresses of all other nodes that you want the node to be able to reach.

When configuring the network directory for a Point-to-Point network, make sure that the IP address you enter in the network directory matches the data in the mapping screens (path name NETXPORT.NI.NIname.MAPPING.mapentry).

For nodes on an X.25 network, the network directory maps the X.25 address key to an IP address to allow a node to communicate within the X.25 network. You must configure a network directory for nodes using X.25.

Planning the Network Directory

There are two theories about how network directories should be planned and configured on a network, as follows:

Centralized network directories.
Decentralized network directories.

The centralized theory requires each node on the internet to have the same network directory. This means that every node in the network must have an entry in the network directory. The advantage to this is that you update the network directory in one place, then copy it to the rest of the world. The disadvantage is that network directories for large internets are going to be large.

The recommended way to create and maintain your network directory using the centralized method is to assign a single node as the central administrative node. You configure the network directory on this node and then copy it to all other nodes on the network. When the network directory is updated, it is updated on the central administrative node, then copied to the other nodes. This procedure decreases the possibility of incompatible directories. You may want to assign a central administrative node for each network or for the entire internet.

The decentralized theory suggests that each network directory be configured individually on each node. The advantage to this is that you can customize the network directory on each node for security purposes using local and global entries. The network directory will also be smaller because it will only contain entries for that particular node. However, updates must be done manually on each node.

Copying and Merging Network Directory Files

The first time you configure the network directory, an entry for all remote IP addresses must be added manually using the NMMGR screens. After the first network directory is configured, you can use the MPE STORE and RESTORE commands to copy the network directory to other nodes. (This is assuming you have adopted the centralized method of network directory maintenance. If you use the decentralized method, you must always use NMMGR to create and maintain the network directory.)

NOTE: The network directory uses a KSAM file pair. Therefore, when copying a directory, be sure to copy both the data file and the key file. The system names the key file automatically using the first six letters of the network directory file name appended with a K. For example, NSDIRK.NET.SYS is the name of the key file associated with the data file NSDIR.NET.SYS.

Once a network directory has been established on each node in the internet, you can set up a job stream to automate network directory updates. The MERGEDIR command is part of a maintenance interface provided primarily to support the updating of directories using a batch job. Using this method, a job or series of jobs can be scheduled at regular intervals to copy and then merge remote directories into the local-system directory. See the MERGEDIR and the MAKESTREAM commands in Using the Node Management Services (NMS) Utilities.

Probe and Probe Proxy

NS 3000 LAN, 100VG-AnyLAN, and 100Base-T NIs with the IEEE 802.3 protocol enabled are able to make use of a proprietary HP protocol called probe. Probe makes it possible for nodes on an NS IEEE 802.3 LAN, 100VG-AnyLAN, and 100Base-T to communicate without a network directory or domain names. A node can determine connection information about a node on the same LAN by sending a multicast probe request out on the network. The target node recognizes its address in the probe request and sends an individually addressed probe reply with the necessary connection information to the requesting node. The probe request/reply mechanism is sufficient to obtain connection requirements within a network.

If the nodes on that LAN are to communicate with other networks, at least one node on the network must have a network directory. The node with the network directory is called a proxy server. By using the probe protocol, a node without a network directory can multicast a request for an internet address from the proxy server. For backup purposes, you should designate at least two nodes to be proxy servers.

Address Resolution Protocol (ARP)

HP e3000 LAN, Token Ring, FDDI, 100VG-AnyLAN, AND 100Base-T NIs are able to make use of a standard protocol called Address Resolution Protocol (ARP). ARP provides IP address to station address resolution. ARP is enabled when the Ethernet protocol or Token Ring is enabled.

Enabling Probe and ARP

With the concurrent configuration of IEEE 802.3 and Ethernet on a network, both the probe and ARP protocols are also enabled. Both protocols broadcast requests to all nodes on the network to resolve the address of a given remote node.

If you disable IEEE 802.3 on a LAN NI, you also disable the probe protocol. Likewise, by disabling Ethernet, you disable the ARP protocol associated with it. You cannot disable both of these protocols simultaneously; at least one must be active to facilitate network communications.

Internetworks

Network Design Questions