Final Project

Routing Protocols

Static Routing

Routing protocols specify communications and routing of data traffic among routers. Of these, static routing is the most basic. With static routing, a routing table in the computer or router must be manually configured to direct communications to other devices on a LAN, or through the default gateway followed by devices on other networks. The most obvious problem with static routing is that the data traffic is confined to one path, and if a link fails then the route needs to be changed or the link itself needs to be re-established immediately. Also, configuring static routes can be tedious and time-consuming if there is a large number of devices that need to be connected to the network, as well as if new devices are to be added frequently, since each device needs to be configured manually in the routing table. Because of this, static routing can often be the least favorable choice for larger networks.

Dynamic Routing Overview

Dynamic routing protocols were designed with the awareness that loss, route changes, and changes in data traffic are often inevitable. Manual configuration of the routing table is minimized with these protocols, and a lot of time is saved because they are designed to change the routing table on their own according to the behavior of the network. They do a great deal of the “thinking” and deciding what to do based on information collected from other routers. This makes them the ideal scheme for larger networks, and there are various ones to match the specific needs of a given network. Protocols can be open or proprietary, which can effect the devices on which they are applicable.

Despite differences among them, dynamic routing protocols make use of four common concepts: Path determination, metrics, convergence, and load balancing. In more detail, metrics are various measures of which routes are the best (e.g. hop count and bandwidth) and putting them in ranks according to practicality. Convergence is when a router has a clear view of routes. The two types of dynamic routing protocols are distance vector and link state.

Distance Vector Protocols

Distance vector protocols are the simpler of the two main types of dynamic routing protocols. Distance vector protocols involve a router that sends the routing table to neighboring routers (those with a 0 hop count) at intervals, and will notify them of any changes. The neighbor routers will decide together which route is best, based on a distance vector, which is often the hop count. Distance vector protocols are considered most practical in smaller networks because they lack the complexity that might be in demand for routing in larger networks. An example of a distance vector protocol is RIP (routing information protocol). RIP uses hop count as the distance vector for deciding the optimal route. RIP is limited, though, and will only allow up to 15 hops in a route. Also, with RIP, sometimes routers will conflict in the decision of which route is the best.

Another example of a distance vector protocol is IGRP (Interior Gateway Routing Protocol), a proprietary protocol created by Cisco, originally as an alternative to RIP. Its improvements over RIP included holddowns, which prevent a bad route from being established again, split horizons, which prevent routers from sending information back to the source router, and poison-reverse updates, which remove bad routes and place them in holddown. These features together were considered effective in preventing routing loops. Various metrics can be used to measure the distance vector in configuring IGRP. However, it is a classful routing protocol which is known to waste IP address space.

Link State Protocols

Link state protocols involve a relationship among neighboring routers that exchange information regarding the status of a route. Instead of periodic sending of the entire routing table, link state protocols will only require notifications to be sent if a change is detected in a route. They also use “Hello” packets to reaffirm communication with neighbor routers. An example of a link state protocol is OSPF (Open Shortest Path First). It is a widely-used open protocol. OSPF is designed to send route updates only upon detecting changes in routes. These change notifications are called LSAs, or link state advertisements. After it detects changes, it will calculate the optimal route based on the shortest path. It also supports variable length subnet masks to make optimal use of IP address space, which is an improvement over IGRP.

Examples in the OSI Model

In the OSI model, routing protocols as a whole are considered layer 3 (network layer) TCP/IP protocols, due to their nature and purpose of finding the best path to forward packets. The network layer is concerned with packet transmission, therefore routing is a good fit for this description. Also, routing tables are constructed at layer 3. However, some protocols work on various layers depending on their functions, especially those that can be identified as applications that use ports.

For example, OSPF operates on the network layer because it uses an IP protocol number as opposed to a port number for its operations. Specifically, OSPF's protocol number is 89. Other example of routing protocols on layer 3 is IGRP and EIGRP, which are IP protocols 9 and 88, respectively. Static routing is configured manually, and therefore is also found on the network layer, because it is extremely simple and does not require a port to operate. While RIP has communications in the network layer because it is uses IP and is responsible for packet forwarding, it also operates in the application layer because it is an application that runs through port 520 UDP. BGP is another routing protocol that also works on the application layer using port 179 TCP.

Interoperability in the OSI Model

The basic job of routing protocols, as stated earlier, is to forward packets through the best path available as defined by whichever protocol is being used. At layer 7, protocols such as RIP and BGP will do the work as applications through their assigned ports. At layer 3, these and the layer 3-specific protocols will manipulate the routing tables if needed and transmit data packets after finding a favorable path.

While data is being transmitted, layer 5 (session layer) is responsible here for maintaining the connection among devices that are transmitting and receiving the data packets. Integrity is maintained for these packets due to layer 4 (transport layer), which will have broken down the data from the session layer into smaller packets and placed them in order. These packets are “translated” in layer 6 (presentation layer) where the data now has consistency and is usable.

Routing also relies highly on lower layers of the OSI model. One of the most important is layer 2 (data link layer) because this is the layer in which errors are detected and handled, frames are created for the packets, checksums are verified to ensure that the data is consistent, and the MAC (physical) address of a device is defined and later mapped to an IP address used for routing.

Strength/Weaknesses of Routing Protocols

One of the main strengths of routing protocols is that they make routing much easier than if only static routing were being used. They save a great deal of time because they decide how the routing table is configured by their own measures, and if there is a link failure, they will find a way around it by measuring another route so that data packets can still make it to their destination. As stated earlier, this can help to eliminate the single point of failure (if one link goes down, there will still be a way to transmit data) and, in turn, the need to immediately restore a link or reconfigure possibly the entire routing table. This is often useful for networks in which static routing can be difficult to maintain. Another strength is that there are various types of protocols to cater to specific types of networks, depending on size among other factors such as use and location.

A weakness of routing protocols is that with increased complexity, there is increased overhead. This can lead to a decline in performance, so it is important to consider which routing method is the most practical for each specific network. Link state protocols are very complex and therefore are more demanding of resources than distance vector protocols. In the smallest of networks, static routing can actually be the best choice because a small number of devices can be easily monitored, and dynamic protocols may be too demanding of resources to make up for the convenience that they were designed to offer. Another weakness that should be considered is that there is not one single type of protocol that is designed to be optimal for all types of networks.

Usefulness over Time

Many of today's routing protocols have been in use for a number of years, and are updated with new versions if they are considered to be outdated. They are designed to be changeable and fit the needs of various types of networks, but a good factor to consider is how networks might continue to change and increase in complexity over the years. New types of routing protocols are certainly a possibility.

Of course, it will always be completely necessary to have a route that is reliable for the transmission of data. With this said, there is little chance that routing protocols will become useless. Routing protocols could, however, change and become more complex. One possibility is the implementation of a universal routing protocol that does not operate in favor of any single type or size of network.

Routers

The main hardware used for routing, is, if course, a router. Routers can be wired or wireless. Various types are used for different purposes and network sizes. A few different types of routers are (provider) edge routers, subscriber edge routers, inter-provider border routers, and core routers. It is important to note that switches can also often be equipped with routing capabilities.

Types of Routers

An edge router is a router that connects various networks together, routing data packets between LANs and larger networks such as WANs, campus networks, and ISPs. A core router only operates within a network and does not provide connection between other networks. An inter-provider border router is a type of router that ISPs use to communicate with other ISPs. Inter-provider routers typically use BGP. This is because BGP provides extreme efficiency for larger routing tables, which is a necessity for ISPs and large networks.

Vendor Offerings

To compare vendor offerings, three different series of routers will be discussed. Each series is from a different vendor but they have similar purposes and specifications. For these devices, routing specifications will be noted. The routers that will be described are the Enterasys S-Series, Juniper Networks M-Series, and Cisco ASR 1000 series. Each series offers routers for various sizes of enterprise as well as service providers.

The first routers that will be discussed are the Enterasys S-Series family of switches that also operate as routers. This family offers a variety of switch routers for different purposes, such as varying enterprise sizes, edge routing, core routing, and data centers. Within this family, there are 3 series: S130, S140, and S150. The S130 and 150 series both provide support for static routing as well as OSPFv2 and RIPv2. These series also support IPv4, IPv6 and multicasting routing. Other features include NAT and LSNAT (load sharing/network address translation), and TWCB which contacts a cache server to help decrease bandwidth use. LSNAT is used for load balancing. Specifically, the S155 devices are able to use BGP which is an efficient protocol for larger routing tables.

The second set of vendor offerings that will be discussed is the Juniper Networks M-Series multiservice edge routers. All M-Series routers have a feature known as a PFE (packet forwarding engine). They support IPv4 and IPv6. They seem to be highly focused on their VPN capabilities and features. These routers support all of the protocols mentioned in the previous offering, but will also support OSPFv3 and IS-IS. Another important feature, NAT, is used in these devices as well.

The last vendor offerings that will be discussed is the Cisco 1000 Series. Devices in the 1000 series contain an integrated firewall and NAT capabilities. For routers in this series intended for service providers, up to 4 million IPv4 and IPv6 routes can be supported. Also, each one can provide support for up to 64,000 subscribers. Because of their nature as edge routers for large enterprises and service providers, BGP is highly recommended and optimized for use with these routers.

Reflection/Views on this Technology

Prior to my research, I was not too familiar with the concept of using protocols for routing. I understood static routing decently well because it was covered in class a few times and was reasonably simple, but it was extremely interesting to read about other approaches to routing and learn about how each of the various types of protocols worked. Even after reading the chapter in the book that covered routing protocols, there was still a lot to learn about them and how they worked. This research was a valuable experience because it helped me to understand routing much more extensively than the book had described.