Design and application of BGP

650) this.width=650; "src=" http://s9.sinaimg.cn/mw690/001VW1Shgy6QHeuDxjid8&690 "width=" 502 "height=" 119 "Name= "image_operate_27391426334650074" alt= "March 14, 2015" title= "March 14, 2015" style= "margin:0px;padding:0px;border:0px; list-style:none;height:2px;width:500px; "/>650" this.width=650; "src=" http://s10.sinaimg.cn/mw690/ 001vw1shgy6qhewuakp59&690 "width=" "height=" 119 "Name=" image_operate_78591426334670765 "alt=" March 14, 2015 " Title= "March 14, 2015" style= "margin:0px;padding:0px;border:0px;list-style:none;"/>

, as the internal router does not run BGP, the edge router runs BGP.

The R1,R5 advertises the loopback interface.

When the R1 1.1.1.1 Ping R5 's 5.5.5.5 loopback, it will be found to be non-pass.

What is the reason for this?

Since the R3 only runs the OSPF protocol and does not run the BGP protocol, only the underlying communication within the zone is guaranteed, and the outside is unable to communicate, so R3 does not learn the R1,R5 route entry, so R3 here produces a routed black hole.

Cheerleading, so how do we solve the problem of routed black holes?

There are four methods listed below, and the fourth method is the focus of today.

1. We can realize the IBGP in as 234, but pay attention to the router configuration under the self-jumping, otherwise IBGP learned the route is not up to. Of course, this is a more cumbersome way, because the router, with command with the hand must be broken ah.

2. We can also establish IBGP in R2 and R3, R3 and R4, establish a partial interconnect within as 234, and then use IBGP as a route reflector. Because the route learned between the default IBGP peer cannot be passed to another pair of IBGP peers, we can only route the reflector.

3. Route redistribution in R2,R4, redistribution of BGP routes into OSPF, but note that when BGP is heavily distributed into the IGP route, the BGP advertised local routes and the routes learned from Ebgp-peer are only re-distributed by default, and routes learned from Ibgp-peer are not re-distributed. If you want to redistribute ibgp-peer routes, it is also very simple, just enter the BGP process and enter the BGP redistribute-internal.

4. The fourth method is what I think of today, it feels convenient, Duang,duang.

Think about it, why does R3 produce a routed black hole? Not because I can't go to 1.1.1.1 and 5.5.5.5, but my next-door brother R2,r4 know, is there a way to let me just drop the bag to my next-door router?

That's MPLS!!! The MPLS label belongs to the 2.5 layer, we can only check the label, do not Anza, so that will not be lost packets. As long as the MPLS IP is turned on in the interface between the R2,R4, the MPLS function is turned on. In this way 1.1.1.1 ping 5.5.5.5, R1 found that the data to R2 lost, R2 received the R4 was found to be lost (because R4 5.5.5.5 the entry to R2), but found 4.4.4.4 again to R3 lost, labeled (IGP learned the route +LDP interaction information), When the packet enters the R3, it discovers that the packet carries the label, then queries the MPLS forwarding publication, then passes to the R4 through the forwarding publication. R4 of course know how to go R5. So the bag is ready to arrive. Come back also the same.

But remember, when r1ping5.5.5.5, be sure to bring the source! Today committed a 2B of the wrong, engaged for a long time have found no pass.

Of course, the fourth method involves MPLS, another day to re-organize the knowledge of MPLS.

Design and application of BGP