CCIE Blog

OSPF virtual links are relatively simple to configure and you normally do not run into too many problem getting them up and working but an odd issue you could run into is when trying to run the virtual link over an interface who’s OSPF cost is maximized (65535 or 0xffff).  The virtual link will not come up if the only interface to reach the other end of the virtual link has a cost that is maximized.  For those of you who have not read RFC 2328 I will quote part of section 15 for you below

Note that a virtual link whose underlying path has cost
greater than hexadecimal 0xffff (the maximum size of an interface
cost in a router-LSA) should be considered inoperational (i.e.,
treated the same as if the path did not exist).

Now you may say why would you ever set the cost to 65535 to begin with?  You may not directly set the cost but you may be asked to use the auto-cost reference-bandwidth command for a task and indirectly set the cost of a transit interface for a virtual-link you created earlier to 65535.  So by solving the auto-cost reference-bandwidth task you broke the virtual-link you created earlier and in turn broke a big portion of your OSPF domain.  In fact now that I think about this issue I am going to write it into version 5 of the R&S material

There are a lot of rumors floating around in regards to diagrams in the R&S CCIE lab.  Cisco officially has said little in regards to this other than the following “the lab document has L1/L2 diagrams for the physical connectivity as well as an IP or topology diagram and an IP Routing diagram”.  This is similar to what we provide in our labs but I would venture to say that they don’t take the time we do to ensure that they look as nice as ours ;)  What Cisco and we do not provide is a true layer 2 “logical” diagram but Cisco and we do provide is a physical diagram of the connections in the lab.  A physical diagram is not the same as a logical layer 2 diagram.  A logical layer 2 diagram will include the VLAN assignments, trunks, EtherChannels, dot1q tunnels, VTP and possibly spanning tree information like root bridges, root ports, designated ports, etc.  The choice to draw out the spanning tree information will really come down to the lab itself.  If there are a lot of tasks that relate to spanning tree or layer 2 traffic engineering (i.e. traffic for VLAN 100 should transit SW3, etc) then adding the spanning tree information will help answer these types of tasks.

The logical layer 3 diagram will be provided BUT the diagram they provide may not have the level of detail you want or need plus you can not write on the diagram they give you.  Technically you can write on it but they will suspend you from the lab for one year ;)  We ALWAYS recommend making your own layer 3 logical diagram.  You should also draw out the diagram for every practice lab you do.  Do not wait until the real lab to draw out your first diagram.  As I have said before you never want to do anything in the CCIE lab for the first time other than get your number

There are two main benefits to making your own logical layer 3 diagram.  First off you will find it is easier to remember what the network looks like when reading the tasks and secondly you will be able to draw and/or take notes on your own diagram.   Smart people fail the lab all the time because they make stupid mistakes in the lab and by drawing out the network you will hopefully lower the chances of making these stupid mistake (i.e. configuring RIPv2 on the wrong interface, applying an ACL inbound on one interface when it should have been outbound on another, configuring a feature on the wrong router, etc).  All it takes is two or three of these little mistakes and you have lost 8 or 9 points in the lab.  We all know that it is hard enough to pass the lab without adding in stupid mistakes into the mix ;)  You will also find tasks related to BGP to be easier to answer when you have a diagram that you can take notes on (i.e. who is peering with who, which exit point to use to reach another AS, etc).  It is possible that when you get into the lab that basic BGP is done for you.  It is normally easier to work on a network that you built from the ground up so working on a network that is 50% complete without first taking the time to discover and document what is already done will be harder.

I am sure someone will comment on this and say, “but I won’t have time to draw out the network in the real lab”.  If this is the case you should not be in the lab in the first place.  If it is taking you the full 8 hours to just configure the network you more than likely will not pass the lab to begin with so taking the 10 minutes to draw out the network is not going to really matter in this case.  The percentage of people who pass the lab while configuring the network for the full 8 hours is slim.  Most people who pass the lab complete the lab within 5.5 or 6.5 hours and have the extra time to do the diagram in the beginning.