Networking on z/OS
Previous topic | Next topic | Contents | Glossary | Contact z/OS | PDF


Routing tables and protocols

Networking on z/OS

There are two different methods for populating a routing table with routes: using static routing, or dynamic routing.

Static routing
Static route is the term applied to any route in a routing table that has been manually coded (entered). For example, when the routing requirements between networks are very simple, routing tables can easily be coded directly into the host to provide all connectivity requirements.

Static routing has limitations when networks become larger. The number of routes can become difficult to manage. Also, networks can change: routers can become unavailable, causing certain routes to be unusable. At the same time, new routes can become available and these must manually be added to the routing table before they can be utilized. To overcome such limitations, dynamic routing can be used.

Dynamic routing
Dynamic routing involves the usage of routing protocols to communicate information about the status of routes and interfaces. z/OS supports two types of dynamic routing protocols: Routing Information Protocol (RIP) and Open Shortest Path First (OSPF).

Routing Information Protocol (RIP)

RIP protocol comes in two versions, RIPv1 and RIPv2. Both protocols require a server (called a daemon) running on the host. This daemon communicates with other hosts running an RIP daemon on the network. Information about the routing tables of each daemon host is exchanged periodically. Routing tables are built based upon information about the network supplied from other routers.

The advantage here is that if a network changes, for whatever reason, the exchange of information among routers allows this change to be communicated. The drawback of RIP is that the routing tables become large very quickly. A large network can require huge routing tables. And, RIP can be slow in recognizing changes in the network. The recognition of changes in a network by a dynamic routing protocol is referred to as convergence.

RIPv1 is seldom used today. RIPv2 has all the functions of RIPv1 plus some improvements. RIPv2 provides a mechanism for authentication of other routers. It provides the ability to multicast, which can reduce network traffic when compared to network broadcasting. RIPv2 also supports variable length network masks.

Most z/OS networks are moving away from using RIPv2 and are instead utilizing OSPF as the dynamic routing protocol.

RIPv2 is defined in RFC 2453.

Open Shortest Path First (OSPF)

OSPF effectively accomplishes the same thing as RIP does: it populates the routing table of a host with routes. It essentially has all the capabilities of RIPv2. However, OSPF is more scalable and configurable than RIP.

In addition, OSPF supports the organization of networks into areas. These areas can be used to limit the amount of information that is required to be moved around an entire internet, yet there is no compromise of connectivity.

From a network route management perspective, OSPF differs significantly from RIP. OSPF exchanges information on the state of links (interfaces) instead of routing information. Link state changes are immediately reported (using a Link State Advertisement). Consequently, network convergence is fast and consistent. In addition, hosts participating in OSPF routing are assigned specific roles (for example, a designated router or an area border router).

The protocol itself is a state-oriented protocol. Interfaces and neighboring routers are always classified as being in a particular state. Events on the network will cause these states to change in a pre-determined way, providing predictability and control to the OSPF routers on the network.

The routing daemon on z/OS (called OMPROUTE) is capable of handling both OSPF and RIP interfaces concurrently.

OSPF is one of the most widely implemented routing protocols. It is defined in RFC 2328.





Copyright IBM Corporation 1990, 2010