Since TCP/IP is a routed protocol, routing is critical to its function. This means that without routing, VoIP won't work either. There are many important things to understand about how routing behavior will affect VoIP services. In this tip, you'll learn routing basics that VoIP managers need to know.
First, routing is basically a function performed by a device to determine where to send a packet. Routing is performed on hosts, such as PCs, servers and IP phones, as well as "intermediate systems," which may often be called routers. On hosts, the configuration is often very simple: a "default gateway" setting, which is actually the IP address of a local router. This gives the host a place to send packets when it doesn't have a better route to the address.
On routers, the configuration is often more complex, as routers normally employ a routing protocol such as OSPF, BGP or RIP to collect information about where addresses are and then calculate a loop-free path through the network to reach them. The process of all the routers in a network choosing the best paths through the network and reaching equilibrium is called "convergence." After a change in the network -- such as a circuit or router failure -- all the routers go through this process again, which we call "reconvergence."
 |
Basic routing protocols
(from Whatis.com) |
There are dozens of routing protocols out there, but these are three you'll hear about very often.
RIP (Routing Information Protocol)
Using RIP, a gateway host (with a router) sends its entire routing table to its closest neighbor host every 30 seconds. The neighbor host in turn will pass the information on to its next neighbor, and so on, until all hosts within the network have the same knowledge of routing paths. RIP uses a hop count as a way to determine network distance.
OSPF (Open Shortest Path First)
Using OSPF, a host that obtains a change to a routing table or detects a change in the network immediately multicasts the information to all other hosts in the network so that all will have the same routing table information. Unlike RIP, the host using OSPF sends only the part that has changed. Rather than simply counting the number of hops, OSPF bases its path descriptions on "link states" that take into account additional network information.
BGP (Border Gateway Protocol)
Unlike RIP and OSPF, BGP is an exterior gateway protocol used for exchanging routing information between gateway hosts in a network of autonomous systems. BGP is often used between gateway hosts on the Internet. The routing table contains a list of known routers, addresses they can reach, and a cost metric associated with the path to each router so that the best available route is chosen. |
|
|
|
|
So that brings us to the important things for voice managers to know about routing. How long does it take to reconverge? And what are the characteristics of the alternate paths?
The time it takes to reconverge is a function of many things, including the routing protocol, selected timer settings, and the number of devices and circuits in the network. Obviously, the more devices and alternate paths there are, the longer it will take the routers to figure everything out. But unless your network is very large (or poorly designed) this is usually not an issue.
The timer settings usually control such variables as how long routes are kept, how often updates are sent, how often "hello" packets are sent, and how long to wait to receive a "hello" packet before declaring that a neighboring router is unavailable. In theory, the lower these numbers are, the faster the network reconverges. But if you set the numbers too low, you can create instability. This happens, for instance, when a very brief hiccup causes a few packets to be dropped. This would normally be transparent to the routing protocols and cause only a few retransmissions by applications, but because your timers have been set so sensitively, these dropped packets cause neighboring routers to break their relationship, and then the whole network reconverges. So the end result is a dropped call instead of a couple of seconds of poor call quality.
Since everyone's network is different, your network team should tune these timers to your network, but you should take care to model how long it takes to restore connectivity after any type of failure.
You also want to understand the characteristics of alternate paths. Assuming your primary path through the network is sized appropriately and configured with QoS and so on, if a given component in that path fails, what is the alternate path? And do you want your voice media to traverse this path? Or would you prefer your voice media to take a special path, like failing over to a local PSTN gateway instead of using a backup WAN circuit?
About the author:
Tom Lancaster, CCIE# 8829 CNX# 1105, is a consultant with 15 years of experience in the networking industry. He is co-author of several books on networking, most recently,CCSP: Secure PIX and Secure VPN Study Guide, published by Sybex.
