Hello Steven,
In essence it will use the interface whose metric to the ASBR is smallest as the outgoing interface.
Yes, you are correct. To put it succintly, in the case the same external type-2 network is advertised by multiple ASBRs, the router will first choose the route with the smallest type-2 metric (which is constant, as you have correctly noted). If there are still multiple choices, OSPF will choose the path with the lowest total metric to the advertising ASBR. If still multiple choices remain even after this second elimination round, OSPF will use all remaining paths and the router will load balance among them.
- Does the router put BOTH routes in the routing table? I am assuming not (I shall go and GNS this shortly)
Yes but only if both routes provide the same distance to the ASBR or ASBRs advertising the external network with the same lowest type-2 metric.
- Is there any way the router can equal cost load balance between the two (afterall the metric IS the same).
Only if multiple routes are installed by OSPF into the routing table. The elimination process is described right above.
If the metric had been type 1 it would have incremented as it passed through the network - meaning that any given router would use the interface whose metric to ASBR is smallest
No, this is not correct. With type-1 metrics, the router will choose the path with the lowest total cost, summing the total distance from itself to the ASBR, plus the starting metric with which the route was redistributed into OSPF. With type-1 metrics, OSPF may choose a route through more faraway ASBR as long as the total cost is the lowest available.
The only place where I can figure that E2 routes are useful is where there are multiple redistribution points. That way you can force the routers to prefer one ASBR over another. Beyond that, I cannot see any reason to using E2 routes over E1 routes - especially in a network with a single redistribution point.
Completely true. With a single redistribution point, the difference between type-1 and type-2 metrics is insignificant. RFC 2328 puts it quite nicely:
OSPF supports two types of external metrics. Type 1 external
metrics are expressed in the same units as OSPF interface cost
(i.e., in terms of the link state metric). Type 2 external
metrics are an order of magnitude larger; any Type 2 metric is
considered greater than the cost of any path internal to the AS.
Use of Type 2 external metrics assumes that routing between
AS'es is the major cost of routing a packet, and eliminates the
need for conversion of external costs to internal link state
metrics.
After typing all of this. It has dawned on me that because of the way OSPF works (link state) the only circumstance in which you would receive "two routes for prefix A" as I state above is if there were 2 redistribution points.
Well, you may have multiple paths to a single ASBR, you may have multiple ASBRs, and you may have multiple paths to multiple ASBRs. You always need to have a mechanism to distinguish which path (or set of paths) is the best. So even in a network with a single ASBR, you may see multiple paths towards it. So even here, you need to choose.
Remember, the type-2 works as follows:
- Choose the ASBR advertising the lowest type-2 metric.
- If there are multiple such ASBRs, select the one that is closest to you.
- If there are multiple such ASBRs, use all of them.
Type-1 works differently:
- Choose the ASBR that provides the least sum of type-1 metric plus the cost to reach it.
- If there are multiple such ASBRs, use all of them.
Feel welcome to ask further!
Best regards,
Peter
P.S.: These rules get more complicated if the ASBRs also advertise the forwarding address in their LSA-5. Don't try to wrap your head around those for the time being.
