Before you install and configure Solstice X.25 9.2, you need to decide how it will be used and how it will interact with other software present on your network.
If you want to connect to a public network, for example one managed by a PTT, you need to provide the network administering body with information about the kind of connection and configuration you would like. Typically, when you contact the administering body, you are sent a network subscription form. Use this to specify what you want. Public networks vary in the services they offer and in the information they require. You may need to answer some or all of the following questions:
Do you want a dial-up line or a leased line? Consider the tariff structure of the public network you are connecting to and how you want to use your network connection.
Do you want a single connection or an MLP connection? You can only use MLP across a public network if you have subscribed to an MLP type line.
What mix of Permanent and Switched Virtual Circuits do you need? A Permanent Virtual Circuit (PVC) is a permanent association between two specified end-points. A Switched Virtual Circuit (SVC) is a temporary association between two endpoints, established for the duration of a call. Compare the tariff structures for PVCs and SVCs before deciding how many of each you need. Some network providers charge a flat rate per month for PVCs and a charge per minute for SVCs, as well as a charge per unit of data on either type of virtual circuit. In general, SVCs tend to be cheaper unless two end-points need to be in permanent communication, but this may not be true for all networks.
What line speed would you like? This is worth investigating carefully, as line speed can have a significant impact on network performance and on the price you will pay for your connection. Consider factors such as the volume of data you expect to transfer, the level of interactive traffic, the type of virtual circuits you plan to have, the expected duration and frequency of connections, the types of modems you have available, and so on.
What is the maximum size of Information Frame (I-Frame) you need? This must be larger than the maximum possible packet size on your network.
How many times should a frame be retransmitted before an error is reported? A large value increases the chance of data getting through correctly. A small value allows fast detection of errors. The value you choose may also depend on whether or not the public network charges for each failure.
What is the maximum packet size you require? In most cases, use of the default value of 128 is appropriate.
What is the maximum window size you require? In most cases, use the default value of 2. A larger window size increases throughput but will probably also increase the cost.
Which throughput class do you require? The throughput class determines the amount of resources allocated to a virtual circuit. It is determined by the line speed. The default value is 2.
Do you need non-standard flow control? If you have chosen non-default packet and window sizes, you may need to use non-standard flow control.
Do you want local or remote packet acknowledgment? Local packet acknowledgment is simpler, so use this, unless a particular application requires remote packet acknowledgment.
If you are connecting to or creating a private network, you need to consider all of the issues described in "Connecting to a Public Network", although cost is not an issue in this case. If you are creating a new network, you may find it simpler to use the default values provided by the Solstice X.25 9.2 software, at least in the beginning. You can fine-tune the configuration later.
This section gives you an idea of how to plan a configuration to run IP over X.25, and to help you find the information you need to do so. Internetworking with TCP/IP: Principles, Protocols, and Architecture, by Douglas Comer, is a useful source of information on IP routing. The SunOS 5.0 manual Administering TCP/IP and UUCP also has information on routing that will help you to plan for your own particular system.
The following is a summary of what you need to do if you plan to use IP routing over X.25:
Obtain and allocate IP addresses.
Every IP network that you intend to connect with other IP networks must have a unique IP network number.
Decide whether to use dynamic or static IP routing. You need to weigh the cost of running a dynamic routing protocol such as RIP across an X.25 connection, against the cost of network downtime caused by out-of-date static routing tables. This decision will be influenced by the size and complexity of your network, as well as by the importance of maintaining connections.
If you decide to use dynamic routing, you need to initiate RIP, the Routing Information Protocol. This is provided in the file in.routed. The contents of the file are broadcast every 30 seconds. You can also use gated, if it is available on your system.
A good compromise between using dynamic and static routing is to configure a static route to a routing gateway that is not attached to the X.25 network. The routing gateway can then handle the dynamic routing of traffic that does not need to cross the X.25 network.
To look at the contents of the routing table, enter netstat --r.
Provide address resolution information, to allow IP addresses to be translated into X.25 addresses, and vice versa. You do this using the IP Interface Configuration window and the Remote Host to X.25 Address Map window. Both windows can be accessed through the Services menu in x25tool. You do not need to provide the address resolution information if you are connecting to the DDN. The DDN uses an automatic address translation mechanism, defined in RFC 1236.