- A virtual community of network engineers
 Home  BookStore  StudyNotes  Links  Archives  StudyRooms  HelpWanted  Discounts  Login
RE: Routing Loops [7:68328] posted 04/30/2003
[Chronological Index] [Thread Index] [Top] [Date Prev][Date Next] [Thread Prev][Thread Next]

Let's see if I can answer this correctly...

The example I can best describe is where two routers connected together
are configured as such:

Router 1 (ISP Core Router) is configured with a static route stating
that w.x.y.z network is reachable via a specific interface.

Router 2 (Customer or ISP Edge Router) is configured with just the
" InterfaceX or IP-address-of-ISP-Core-Router)"
It may also have its own local/directly connected networks defined in
RIP/OSPF/EIGRP.  If the interface for one of those directly connected
networks (off of another logical/physical interface other than the
upstream ISP/Peer Router) goes down, the route is removed from the
routing table.

Router1 sends a packet destined for router 2 (because that's where its
routing table says it goes due to the static route).
Router2 looks at the packet, checks its route table, doesn't fine an
entry for the destination network, and responds with a "I don't know
where this goes" and sends it out its interface that is configured for
the InterfaceX or IP-address-of-ISP-Core-Router entry
(also known as its default route).

This creates a ping-pong affect of Router1 saying "hey, this packet
belongs to you" and Router2 saying "dude, I have no idea where it goes-
you figure it out".

And thus a routing loop is born.

An example of how I have experienced this is a customer router goes down
due to power failure.  The ISP Edge router drops the route to the client
network from its table.  The ISP Core Router (maintaining a BGP entry
for that Customer Network) gets an HTTP packet destined for the
customer's Web Server (which was hosted by the customer themselves).
The Core router forwards it to the Edge Router, and the Edge Router
sends it back to the Core router because it doesn't know where to
deliver it.

When Joe Blow decides to perform a TraceRoute from his computer to the
WebServer he was trying to reach, he gets something like the following:

--First 8 Hops snipped for privacy sake--

9  58 ms 55 ms 59 ms []
10 59 ms 59 ms 59 ms []
11 65 ms 66 ms 64 ms []
12 66 ms 65 ms 66 ms []
13 89 ms 89 ms 89 ms []
14 88 ms 89 ms 88 ms []
15 102 ms 105 ms 105 ms
16 103 ms 102 ms 101 ms []
17 104 ms 102 ms 106 ms
18 106 ms 106 ms 104 ms
19 109 ms 107 ms 105 ms []
20 106 ms 106 ms 104 ms
21 109 ms 107 ms 105 ms []
22 106 ms 106 ms 104 ms
23 109 ms 107 ms 105 ms []
24 106 ms 106 ms 104 ms
25 109 ms 107 ms 105 ms []
26 106 ms 106 ms 104 ms
27 109 ms 107 ms 105 ms []
28 106 ms 106 ms 104 ms
29 109 ms 107 ms 105 ms []
30 106 ms 106 ms 104 ms

Now, this is just for example... but it is like a simulation of the
Oracle.Com Public Network going off-line, and thus unreachable.

You can see that from hop 19 on, a routing loop is in effect (or at
least this is what it would look like from a traceroute.

-----Original Message-----
From: ccnp ccnp2002 [mailto:wakumu2002@xxxxxxxxx] 
Sent: Wednesday, April 30, 2003 8:19 AM
To: cisco@xxxxxxxxxxxxxx
Subject: Routing Loops [7:68328]

Hello All,

Now, another question, hopefully simpler.

I have read about it for more than one year now; I can visualize a loop
in a
switched network and how STP takes care of that.

But I have never visualized a routing loop; if I was asked to define it
to a
novice, I guess I would have difficulty (being a novice myself).

In very simple terms, what is a routing loop?


Message Posted at:
FAQ, list archives, and subscription info:
Report misconduct and Nondisclosure violations to abuse@xxxxxxxxxxxxxx