RIP: Routing Information Protocol

Learning objective

Explain how RIP exchanges routes by hop count, prevents loops with split horizon and triggered updates, and where its RFC 2453 packet format and limits show up in practice.

What RIP is

RIP (Routing Information Protocol) is the oldest interior gateway protocol still seen in production and lab networks. RIPv1 was standardized in RFC 1058 (1988); the classless version in wide use today, RIPv2, is standardized in RFC 2453 (1998). RIP is a distance-vector protocol: each router tells its neighbors only "here is my distance to each network I know," using a single metric — hop count — the number of routers between here and the destination. RIP runs over UDP port 520, and RIPv2 sends its periodic updates to the multicast address 224.0.0.9 (RIPv1 used broadcast).

RIP packet format

An RFC 2453 RIPv2 message starts with a command field (1 = Request, 2 = Response), a version field, and then up to 25 route entries. Each route entry carries an address family identifier, the destination IP address, a subnet mask (the field RIPv1 lacked, forcing it to assume classful boundaries), a next hop address, and the metric itself — a single byte, since RIP's whole distance space only needs to represent 0 through 16.

How RIP updates work

RIP routers send their full routing table as a periodic Response roughly every 30 seconds. If a route hasn't been reheard within 180 seconds, it is marked invalid and advertised with a poisoned metric of 16; if it stays unreachable, it is flushed from the table entirely around 240 seconds after it was last valid. Beyond the periodic cycle, RIP also sends triggered updates: the instant a route changes — a link fails, or a better route appears — the affected router advertises the change immediately, without waiting for its next 30-second cycle.

Preventing routing loops

A pure distance-vector protocol is naturally vulnerable to routing loops: if two routers each believe the other has a valid path to a now-dead network, they can bounce updates back and forth, each incrementing the hop count, until they "count to infinity." RIP caps this at hop count 16 (defined as unreachable), and adds two loop-prevention rules to stop it happening in the first place:

• Split horizon: never advertise a route back out the same interface it was learned on.
• Split horizon with poisoned reverse: go further — advertise that route back out the same interface, but with a metric of 16, explicitly telling the neighbor "don't route through me for this."

Interactive RIP exchange

Step through two RIP routers exchanging a route, applying split horizon with poisoned reverse, and reacting instantly to a link failure with a triggered update.

RIP's limits and RIPng

Hop count treats every link as equal, so a 15-hop path of gigabit Ethernet loses to a 2-hop path of dial-up — RIP has no concept of bandwidth. A maximum usable hop count of 15 also puts a hard ceiling on how large a RIP network can grow. RIPv1's lack of a subnet mask in its route entries meant it could not support Variable Length Subnet Masking (VLSM) or discontiguous subnets; RIPv2 fixed this by carrying a mask with every route. For IPv6, RIPng (RFC 2080) adapts the same distance-vector, hop-count design to 128-bit addresses, sent over UDP port 521 to the multicast group FF02::9.