Introduction
The purpose of this document is to critically evaluate two complex Interior Gateway Protocols (IGP) and provide a balanced and objective review on their strengths, weaknesses, operational requirements and discuss their commercial implementations. The chosen protocols are Enhanced Interior Gateway Routing Protocol (EIGRP) and Open Shortest Path First (OSPF). These protocols are selected due to their dominance in today’s enterprise computer networks and the fact that each of the protocols represents a different major protocol class. EIGRP a distance-vector protocol and OSPF representing a link-state protocol class.
Both of the protocols share some similarities in their operation such as being classless therefore supporting CIDR/VLSM as well as being able to work with IPv4/IPv6 address families (OSPFv3 and EIGRP for IPv6) and ensure loop-free routing. However OSPF and EIGRP are very diverse protocols each with its own strengths and weaknesses.
Operational strengths
Even though both of the protocols offer fast convergence times, number of simulation tests reveal that EIGRP has faster convergence times and lower bandwidth consumption in comparison to OSPF it can also take the link bandwidth into account and adjust the volume of the updates sent. Simulations carried out by Thorenoor, S. (2010) using OPNET Modeller reveal that EIGRP outperforms OSPF in terms of network convergence duration, protocol traffic, CPU and network bandwidth utilisation as well as throughput and queuing delay (the method of testing used by Thorenoor, S. is questionable, but both protocols were tested under the same method).
Bandwidth utilisation tests carried out by Wijaya, C. (2011) shows that EIGRP has a lower bandwidth utilisation when routing information exchanged and quicker convergence times:
“The number of packet loss is smaller when using EIGRP as compared with OSPF. Whether it’s using an IPv4 addressing or IPv6 addressing. This is because EIGRP uses DUAL algorithm that includes a successor when a route is lost. By using EIGRP, the packet loss ranging from 1 to 3. Meanwhile, when using OSPF, packet loss ranging from 2 to 4.”
OSPF being a link-state protocol maintains a LSDB – a map of the whole network topology to ensure loop free routing and improve convergence times in the event of a link failure.
In contrast to OSPF, EIGRP does not store the whole topology, but holds a route to the feasible successor router and in the event of a link failure it would switch to the feasible successor route resulting in a rapid connectivity restoration.
OSPF can support route summarisation on the border routers only, while EIGRP offers more flexibility by supporting per interface route summarisation this results in reduced resource utilisation and faster convergence. In order to improve routing protocol scalability both protocols support stub areas/routers (even though the stub areas/routers work in a different fashion), OSPF areas can be configured as stub, totally stubby, not-so-stubby or totally not-so-stubby. EIGRP has more options such as receive-only, leak-map, connected, static, summary and redistributed, all of which can be combined except the receive-only option. To further increase scalability OSPF uses areas to create 2 tier network hierarchy which helps to logically sub-divide the network reducing LSA flooding and LSDB size thus saving resources and making the protocol highly scalable.
EIGRP is unique by its use of Protocol-dependent Modules (PDM) which allow the protocol to support not only IP but also various other network layer protocols like IPX, AppleTalk and adapt to future’s changes.
Both protocols allow equal cost load balancing, EIGRP also supports unequal cost load balancing.
Weaknesses and known issues
OSPF weaknesses include higher CPU, memory and bandwidth utilisation, higher convergence times compared to EIGRP, being unable to support other protocols than IP/IPv6, being complex to configure especially when number of areas grow and the fact that summarisation can only be done on the border routers. Among the EIGRP weaknesses is its lack of interoperability with other vendors and Stuck-In-Active issue, even though mitigated by introducing SIA-query and SIA-reply messages still occasionally causing unnecessary termination of neighbour adjacencies when there are unreliable links.
Operational requirements
The OSPF protocol was developed by the Internet Engineering Task Force (IETF) as a non-proprietary vendor-independent routing protocol this allow the protocol to be used by any vendor and enable customers to maintain a multivendor network environment. EIGRP unlike OSPF is a Cisco proprietary protocol therefore works only with Cisco devices, the protocol has been partially opened by Cisco for the interoperability reasons, but OSPF remains a more attractive option for large scale network applications due to its multivendor support and high scalability.
Link-state protocols generally require more CPU processing power and memory to operate compared to the distance-vector protocols due to the need of maintaining its local LSDB and run SPF algorithm every time there is a change in the local area to re-calculate SPF Tree. Since OSPF only supports route summarisation at the area border, all the intra-area routes will be held in the database. Flooding of the SLAs can cause congestions on the slower links, but OSPF DR/BDR router is used to minimise the number of updates sent and acts as a central distribution point for LSAs in an area.
Wijaya, C. (2011) points out that the number of total packets sent during the routing information exchange is lower when using EIGRP:
In contrast EIGRP requires less CPU, memory and link bandwidth resources due to the fact that EIGRP sends only bounded and triggered updates and does not store the whole network topology in its database. EIGRP maintains a neighbour table and installs a feasible successor route instead. It also sends partial updates in the event of a change to reduce bandwidth utilisation and is capable of capping the amount of updates it sends to prevent congestions.
Commercial implementation
OSPF is one of the most popular IGP in today’s computer networks. Being an open standard protocol means that it will be preferred in a large scale multi-vendor networks due to its interoperability. The hierarchical structure, route summarisation and stub options makes the protocol very scalable and suitable for all sizes of networks.
EIGRP is fast and not as resource demanding as OSPF, it offers high flexibility by enabling per-interface route summarisation and PDMs for various today’s and future’s protocols. EIGRP supports stub networks and is scalable, easy to configure and low overhead. The fact that it is Cisco proprietary protocol can make the protocol look less attractive in a large multivendor network environments.
References
Wijaya, C. (2011) Performance Analysis of Dynamic Routing Protocol EIGRP and OSPF in IPv4 and IPv6 Network [Online], IEEE. Available at http://ieeexplore.ieee.org.libezproxy.open.ac.uk/stamp/stamp.jsp?tp=&arnumber=6141697 (Accessed 4 July 2016)Thorenoor, S. (2010) Dynamic Routing Protocol implementation decision between EIGRP, OSPF and RIP based on Technical Background Using OPNET Modeler [Online], IEEE. Available at http://ieeexplore.ieee.org.libezproxy.open.ac.uk/stamp/stamp.jsp?tp=&arnumber=5474509 (Accessed 4 July 2016)