D-Link DES-3326SR Layer 3 Switch
shortest path back to the multicast source, or that have no active multicast group members. A ‘graft’ message is added that
allows a previously pruned branch of the multicast delivery tree to be reactivated. This allows for lower latency when a leaf
router adds a new member to a multicast membership group. Graft messages are forwarded one hop (one router) back at a
time toward a multicast source until they reach a router that is on an active branch of the multicast delivery tree.
If there is more than one multicast router on a network, the one that has the shortest path back to the multicast source is
elected to forward multicast packets from that source. All other routers will discard multicast packets from that source. If
two multicast routers on a network have the same distance back to a multicast source, the router with the lowest IP address is
elected.
There are two protocols in Protocol Independent Multicast (PIM), Protocol Independent Multicast-Dense Mode (PIM-DM)
which is used when the multicast destinations are closely spaced, and Protocol Independent Multicast-Sparse Mode (PIM-
SM) which is used when the multicast destinations are spaced further apart. PIM-DM is most commonly implemented in an
intranetwork (LAN) where the distance between users is minimal.
The Routing Information Protocol is a distance-vector routing protocol. There are two types of network devices running RIP
– active and passive. Active devices advertise their routes to others through RIP messages, while passive devices listen to
these messages. Both active and passive routers update their routing tables based upon RIP messages that active routers
exchange. Only routers can run RIP in the active mode.
RIP measures distance by an integer count of the number of hops from one network to another. A router is one hop from a
directly connected network, two hops from a network that can be reached through a router, etc. The more routers between a
source and a destination, the greater the RIP distance (or hop count).
There are a few rules to the routing table update process that help to improve performance and stability. A router will not
replace a route with a newly learned one if the new route has the same hop count (sometimes referred to as ‘cost’). So
learned routes are retained until a new route with a lower hop count is learned.
When learned routes are entered into the routing table, a timer is started. This timer is restarted every time this route is
advertised. If the route is not advertised for a period of time (usually 180 seconds), the route is removed from the routing
table.
RIP does not have an explicit method to detect routing loops. Many RIP implementations include an authorization
mechanism (a password) to prevent a router from learning erroneous routes from unauthorized routers.
To maximize stability, the hop count RIP uses to measure distance must have a low maximum value. Infinity (that is, the
network is unreachable) is defined as 16 hops. In other words, if a network is more than 16 routers from the source, the local
router will consider the network unreachable.
DVMRP also supports tunnel interfaces, where two multicast routers are connected through a router that cannot process
multicast packets. This allows multicast packets to cross networks with routers that are not multicast-aware.
Protocol-Independent Multicast – Dense Mode
Routing Protocols
Routing Information Protocol (RIP)
Every 30 seconds, a router running RIP broadcasts a routing update containing a set of pairs of network addresses and a
distance (represented by the number of hops or routers between the advertising router and the remote network). So, the
vector is the network address and the distance is measured by the number of routers between the local router and the remote
network.
RIP can also be slow to converge (to remove inconsistent, unreachable or looped routes from the routing table) because RIP
messages propagate relatively slowly through a network.
Slow convergence can be solved by using split horizon update, where a router does not propagate information about a route
back to the interface on which it was received. This reduces the probability of forming transient routing loops.
Hold down can be used to force a router to ignore new route updates for a period of time (usually 60 seconds) after a new
route update has been received. This allows all routers on the network to receive the message.
A router can ‘poison reverse’ a route by adding an infinite (16) hop count to a route’s advertisement. This is usually used in
conjunction with triggered updates, which force a router to send an immediate broadcast when an update of an unreachable
network is received.
206
Содержание DES-3326SR
Страница 240: ......
