Chin Blog

拓樸：

設定：

★★★請注意，R2與R3沒有建立iBGP關係！★★★

拓樸中的紅字是BGP路由反射的主要設定，反射前，分別是R1、R2與WAN1的資訊。

R1路由表有11.11.11.11 / 32往2.2.2.2(R2)、22.22.22.22 / 32往3.3.3.3(R3)的路由，這2筆分別經由R2、R3傳過來，能從AD值200看出來使用iBGP而非EIGRP。

R2路由表有11.11.11.11/ 32 往20.0.0.1(WAN1)的路由，它從WAN1傳過來，能從AD值20看出來使用eBGP；這時候R2「show ip bgp neighbor 20.0.0.1 advertised-routes」是空的，指令意義是R2通告給鄰居20.0.0.1(WAN1)哪些BGP路徑。

目前WAN1路由表僅有直連網路(C：connected和L：local)；「sh ip bgp」的輸出，WAN1知道11.11.11.11 / 32的BGP路由，並且它的下一站是0.0.0.0，代表是自己產生的，因為這命令「network 11.11.11.11 mask 255.255.255.255」；在WAN1上「show ip bgp neighbor 20.0.0.2 advertised-routes」能看到WAN1通告給鄰居20.0.0.2(R2)哪些BGP路徑。

然後是反射後，R1、R2、與WAN1的資訊。

你能看到R1對R2和R3做路由反射，R1會非常快與R2、R3重建iBGP關係，並且「show ip bgp 11.11.11.11 / 32」的輸出，其中一行寫到「Received from a RR-Client」，RR是Route Reflector的縮寫。

R2路由表有22.22.22.22 / 32往3.3.3.3(R3)的路由，「show ip bgp neighbor 20.0.0.1 advertised-routes」的輸出，得知R2通告給鄰居20.0.0.1(WAN1)這筆22.22.22.22 / 32的路由。

WAN1路由表有22.22.22.22 / 32往20.0.0.2(R2)的路由，「show ip bgp」的輸出，WAN1知道22.22.22.22 / 32的路由，下一站是20.0.0.2(R2)。

最後結論是R1做為BGP反射器、R1是鏡子，反射路由給iBGP，然後R3與WAN2反之亦然，並且日後擴展分公司，增加R4、R5也能夠反射，並不僅限2台路由器；其實這個技術能應用在很多地方，例如MPLS-VPN、EIGRP通過訊框中繼廣域網路並關閉水平分割可能產生的路由迴圈問題(尤其是多點子介面)。

Topology:
[Image]
configuration:
[Image]

★★★To notice, R2 and R3 aren't established iBGP relationship!★★★

Red words are primary BGP route reflector configuration in topology. Not yet to reflect, those are R1, R2 and WAN1 information respectively.
[R1-1]
[R2-1]
[WAN-1]
R1's route table installs 11.11.11.11 / 32 forward to 2.2.2.2(R2), 22.22.22.22 / 32 forward to 3.3.3.3(R3). They were propagated by R2, R3. Routing protocol used iBGP not EIGRP because AD value is 200.
R2's route table installs 11.11.11.11 / 32 forward to 20.0.0.1(WAN1). It was propagated by WAN1. Routing protocol used eBGP because AD value is 20.
「show ip bgp neighbor 20.0.0.1 advertised-routes」is empty on R2. The command is R2 advertise some BGP routes to 20.0.0.1(WAN1).

WAN1's route table installs direct connect network(C:connected and L:local) only currently. Output of 「sh ip bgp」, WAN1 knows 11.11.11.11 / 32 route for BGP, and it's next hop is 0.0.0.0. It is generated by self, because command 「network 11.11.11.11 mask 255.255.255.255」.
You can see WAN1 advertised some routes to 20.0.0.2(R2) after command「show ip bgp neighbor 20.0.0.2 advertised-routes」 on WAN1.

And then is reflected, those are R1, R2 and WAN1 information respectively.

[R1-2]
[R2-2]
[WAN1-2]
R1 is route-reflector with R2 and R3. R1 will reestablish iBGP relationship with R2 and R3 soon. And output of 「show ip bgp 11.11.11.11 / 32」, some row printed 「Received from a RR-Client」. RR is Route Reflector abbreviation.
R2's route table installs 22.22.22.22 / 32 forward to 3.3.3.3(R3). Output of 「show ip bgp neighbor 20.0.0.1 advertised-routes」, R2 advertised route 22.22.22.22 / 32 to neighbor 20.0.0.1(WAN1).
WAN1's route table installs 22.22.22.22 / 32 forward to 20.0.0.2(R2). Output of 「show ip bgp」, WAN1 knows route of 22.22.22.22 / 32, it's next hop is 20.0.0.2(R2).
Last conclusion, R1 be a BGP route-reflector, it is mirror. Router can reflects routes to iBGP. In topology, R3 and WAN2 are vice versa. And based scalability to expand branches in future, add R4, R5 can reflect also. Route reflector technology isn't only two routers.
In fact, the technology can deploy in wide. Such as MPLS-VPN, EIGRP over frame-relay WAN. EIGRP over frame-relay network to disable split-horizon under interface maybe generate route loop problem.(It is multipoint interface espically.)