Hello and welcome to this article in our Network Reliability Series. In the last post, we started examining Layer 3 redundancy technologies and we took a look active/standby and active/active redundancy technologies. We also introduced route redundancy technologies; you can review the post here. In this new post, we would start right from where we left off by examining the inherent redundancy features in Layer 3 routing. So let’s get right to it.
Active Standby Redundancy
In routing, individual next-hop forwarding decisions are made by routers based on the information available to them. The implementation might vary with different routing protocols but the redundancy concept is consistent among them. Let’s consider static routing.
With static routes, the routes are manually entered using the IP route command. For active standby redundancy in static routing, an administrator can force an alternate route to be a backup route by increasing its administrative distance (by default, the AD of a static route is 1). When the AD of a static route is increased to make it less preferable than other static (or even dynamic) routes, this is called floating static routes. Let’s take an example:
In the diagram, router A has two upstream routers (routers B & C) that can be used to reach 192.168.12.0/24. If the path through router B is preferred, then we can increase the administrative distance of the path through router C. To do this, we just add an extra parameter at the end of the IP route command. So on router A, we would have;
ip route 192.168.12.0 255.255.255.0 192.168.1.2
ip route 192.168.12.0 255.255.255.0 192.168.2.2 5
The configuration above increases the AD of the path through 192.168.2.2 (router C) to 5 (from the default of 1). This causes the route to be less preferred and as such, it would not be installed in the routing table. So where is this backup route kept? Well, since it was manually configured by and administrator, then it would remain in the configuration file.
When does the backup route kick in? This is usually the tricky part with static routing protocols since both routes are stored in the configuration file and the primary/active route would always be preferred. Well, one thing to consider is that a route is only installed in a routing table when its next hop is reachable. So the route through 192.168.1.2 would remain in the routing table until 192.168.1.2 becomes unreachable. This makes sense but it can become an undesirable feature when something goes wrong with the upstream router but does not affect the reachability of the next hop address. For example, if interface Fa0/1 of router B goes down, then the 192.168.12.0/24 network would be unreachable but the route would still installed as the primary route. This can be fixed with route tracking.
With route tracking, specific attributes can be tracked with the influence which static route(s) get installed, despite the administrative distance of the routes. There are many attributes that can be tracked and we cannot cover all of them in this article. Thankfully, I’ve got you covered. You can read more about static route tracking here and here.
Active/Standby redundancy in dynamic routes.
With dynamic routes, active/standby redundancy is inherently designed in the operation of the routing protocols. The concept of redundancy is closely linked with the concept of metrics in routing protocols. Let’s examine the redundancy features in RIP, EIGRP and OSPF.
Routing Information Protocol: In RIP, a basic metric of hop count is used. When updates are received, if two updates are received about the same route (same network subnet), the route with the lower hop count is installed in the routing table. Since RIP is a distance vector routing protocol (and an old one at that), the backup routes are not stored, but once the RIP updates are sent again, if a route now has a lower metric, that route is now considered as the primary and installed in the routing table.
Enhanced Interior Gateway Routing Protocol: With EIGRP, there are a lot of concepts that come together to determine redundancy. The two major concepts are the complex EIGRP composite metric (I hope to write about that sometime!) and the feasibility condition. However, the concept of active/standby redundancy can be summarized into two words: Topology Table. The topology table contains the routes received from the EIGRP neighbors. When multiple updates are received from the same route, the route with the lowest metric is selected as the primary route and installed in the routing table. This route is also called the successor route. To determine the routes that would be backups (also called feasible successor), the feasibility condition would be checked. The feasibility condition states that the reported/advertised distance of an update must be less than the feasible distance of the successor route. The backup routes are stored in the topology table and this makes recovery fast when the primary route fails.Open Shortest Path First protocol: OSPF is a link state routing protocol that keeps link state updates in a link state database. The best path to a particular destination is computed using the Djikstra’s shortest path first algorithm. The best path is installed in the routing table. When an alternate path is needed, the SPF is run again and the best path is selected again. This is the standard behavior of OSPF with Active/Standby redundancy. In some cases (incremental SPF and partial SPF), the full SPF algorithm may not be recomputed but these cases are beyond the scope of this article.
To summarize Layer 3 active/standby route redundancy, I have created the table below to highlight the key features of all the routing protocols.
A/S Redundancy Feature
Routing Information protocol
Backup routes are selected during next update
Enhanced Interior Gateway Routing Protocol
Feasible successors are stored in topology table
Open Shortest Path First
Shortest path algorithm is recomputed after every topology change
Active/Active Redundancy in Layer 3 routing
For active-active redundancy in IP routing, the goal for network administrators is to get multiple routes installed in the routing table. So how can multiple routes get installed in the routing table? They must have the same administrative distance and metric. This means that by default, routes from multiple routing protocols (for instance, EIGRP and OSPF) cannot get installed in the routing table. This is because the route from the routing protocol with a lower administrative (in this case, OSPF) distance would always be installed. So how do we achieve active/active redundancy with these protocols?
Static Routing: This is the easiest to achieve. Instead of floating the backup route with a higher administrative distance, the two routes would just be entered with the default administrative distance and they would both be installed in the routing table.
RIP: If updates are received the same hop count, then both of them would be installed in the routing table. Network administrators can influence the hop counts using “offset-lists” to increase the hop count of the active route so that they can be the same with the backup.
EIGRP: With the composite metric of EIGRP (it uses bandwidth, delay, reliability and load); it is difficult to have the same metric from routes that do not have a symmetrical path. Again, administrators can influence the metric by either changing one of the component parameters or just using the “offset-list” to increase the metric. However, neither of these approaches is scalable in the long run. An alternative method offered by EIGRP is unequal cost load balancing. We would explore this later in the series.
OSPF: The OSPF metric is cost, and this is a direct function of the bandwidth of the links involved. Because OSPF is a link state protocol, network administrators have to make extra considerations of the impact of changes they make. The OSPF cost of an interface can be changed using the “ip ospf cost” command. Unlike distance vector protocols, the offset-list command is not available in OSPF. This is because each OSPF router re-calculates its best paths based on the LSAs in its database.
Whew! A lot of concepts were covered in this article. We started by looking at active-standby redundancy in static and dynamic routing. We also identified the individual properties of each routing protocol that affects their implementation of redundancy. Similarly, we explored active-active redundancy and we looked at how we can influence routing protocols so they can have more than one best route to be installed in the routing table.
In the next article, we would consider how redundant routes with equal costs are treated from a load sharing perspective. We would also explore unequal cost load balancing and review the effects of summary routes on layer 3 redundancy. Thank you so much for reading and I look forward to sharing the next article with you soon. Until then, keep building redundant networks.