of 19
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.

BGP Routing. Agenda. L47 - BGP Routing. BGP Internals BGP Session Topologies Synchronization with IGP Fail-over Handling

Category:

Philosophy

Publish on:

Views: 4 | Pages: 19

Extension: PDF | Download: 0

Share
Description
BGP Routing The BGP Routing Principles and Route Decisions based on AS-Path BGP Routing, v4.4 1 Agenda BGP Internals BGP Session Topologies Synchronization with IGP Fail-over Handling BGP Routing, v4.4
Transcript
BGP Routing The BGP Routing Principles and Route Decisions based on AS-Path BGP Routing, v4.4 1 Agenda BGP Internals BGP Session Topologies Synchronization with IGP Fail-over Handling BGP Routing, v4.4 2 Page 47-1 BGP Routing and BGP Policy in a simple topology of AS s routing policy is reduced to a minimal function demonstrated in example 1 and 2 (BGP-4 Fundamentals module) a BGP router can decide only which networks within the own AS should be announced to external BGP neighbors and which learned networks should be advertised into the own AS no route decision must be taken in a complex topology of AS s routing policy is necessary to decide which routes should be propagated to other peers BGP policy based on agreements between AS s in case of several paths to same destination route decision (= selecting the best path) is necessary BGP Routing, v4.4 3 BGP Routing Information Base in a meshed topology of AS s routing decisions are necessary to maintain network connectivity across AS s in case of topology changes e.g. physical link between two AS s is broken to select the best path in case of several paths to same destination in order to handle routing policy and route decision BGP routes are stored in three conceptual Routing Information Bases (RIBs) within a BGP router Adj-RIBs-In, Loc-RIB, Adj-RIBs-Out BGP Routing, v4.4 4 Page 47-2 BGP Routing Information Bases Adj-RIBs-In store routing information that has been learned by inbound Update messages; there contents represent routes that are available as an input to the decision process an optional Input Policy Engine can filter routes or manipulate their attributes (policy decision) before this routing information is given to the route decision process Loc-RIB contains the local routing information that the BGP router has selected after applying its local input policies and route decisions an optional Output Policy Engine can filter routes or manipulate their attributes before this information is given to peers Adj-RIBs-Out contains the routing information that the BGP router has selected for advertisement to its peers BGP Routing, v4.4 5 Model of BGP Process Incoming BGP Updates from peers Outgoing BGP Updates to peers Input Policy Engine Adjacency RIBs out sh ip bgp (Cisco ) Adjacency RIBs IN Output Policy Engine sh ip route (Cisco) BGP Routing Decision Process IP Routing Table Local RIB sh ip bgp (Cisco * ) BGP Routing, v4.4 6 Page 47-3 BGP Routing - Path Vector Protocol with the lack of an Input and Output Policy Engine routing decisions alone will control propagation of BGP routes to peers without special assumptions about use of attributes only AS_Path, Next_Hop and Origin (the mandatory attributes) are used for routing decisions only path vector protocol every routing update (BGP route) contains full list of transit networks (AS_Path) handling very similar to distance vector algorithm e.g. split horizon, hop metrics BGP Routing, v4.4 7 Path Vector Protocol Details 1 basic algorithm of path vector routing when receiving an update with own AS number contained in the AS_Path, the BGP route will be refused loop protection when an update is received, the path will be compared to the current best path to this destination if the new path is better (e.g. shorter) than the old path the BGP routing table is modified to include the new path corresponding updates are sent to BGP neighbors implementations might include extra features BGP Routing, v4.4 8 Page 47-4 Path Vector Protocol Details 2 only the best paths are forwarded to the neighboring ASes same concept as in distance vector protocols however, if this path is broken the next best path should be advertised the BGP routing table is also modified if a new path is received from the currently selected neighbor for that destination if a the currently selected neighbors withdraw a route in this case, the last advertisements from all other neighbors are compared and the best path is selected BGP Routing, v4.4 9 Agenda BGP Internals BGP Session Topologies Synchronization with IGP Fail-over Handling BGP Routing, v Page 47-5 BGP Meshed Topology (example 3) AS internal Router running IGP only BGP Border Router BGP Routing, v External Internal BGP Sessions path information between different AS s is transported by BGP Update message external BGP (EBGP) sessions direct physical link is necessary for simplicity only, not a strict requirement to eliminate the need for yet another DMZ routing protocol remark: some implementations support non-directly connected EBGP peers (e.g. Cisco s ebgp multi-hop feature) in order to pass path information within the same AS to other BGP routers internal BGP (IBGP) sessions are necessary remark: IGP protocols might not be able to pass AS numbers or any other path attributes BGP Routing, v Page 47-6 BGP Internal - External Sessions IBGP session EBGP session R7 AS 4 BGP Routing, v External Internal BGP Sessions different behavior concerning routing updates update received on an internal connection will not be propagated to other BGP routers of same AS basic BGP loop avoidance does not work when AS number is the same through multiple hops! therefore internal BGP routers must be fully meshed IBGP sessions to every other BGP router of same AS if an external update is received and propagated via an internal BGP connection the Next_Hop will be that of the BGP router which originates this external update external router (IP address of this router) must be reachable by IGP or some other means from that internal router recursive lookup of routing table is necessary to find the real next hop in any way, if next hop is not directly connected interface BGP Routing, v Page 47-7 BGP Update (1) Adj-RIBs-In, Local-RIB BGP Table 1.), AS1, 1.) means best BGP route, which is stored in the Local-RIB (AS1), (AS1), BGP Table 1.), AS1, BGP Routing, v BGP Update (2) BGP Table 1.), AS1, (I) means: learned via internal BGP Session BGP Table 1.), AS1, (I) BGP Table 1.), AS1, BGP Routing, v Page 47-8 IGP Routing Update into AS2 (3) BGP Table 1.), AS1, BGP Table 1.), AS1, (I) that means the BGP router imports this route into its IGP process and IGP propagates it to all routers (within this AS) IGP Routing Update BGP Table 1.), AS1, BGP Routing, v Agenda BGP Internals BGP Session Topologies Synchronization with IGP Fail-over Handling BGP Routing, v Page 47-9 BGP and IGP for an internal IGP-only router in order to achieve reachability of Net-ID s of other AS s default routing within an AS towards BGP Border or route redistribution from BGP into IGP maybe a problem in case of carrying all Internet routes rules for redistribution only the best BGP route will be installed in the BGProuters IP-forwarding (routing) table only routes learned via BGP external sessions are redistributed into IGP remark: Cisco-IOS default filter behavior BGP Routing, v BGP and IGP Synchronization if an BGP router receives a route via internal BGP, it must wait until this route is reachable via IGP too before this route could be announced to an external BGP peer this is called BGP synchronization reason: if router would propagate this route earlier, the AS would get traffic for that destination but this traffic could not be passed through the own AS BGP Routing, v Page 47-10 BGP Update after BGP Synchronization (4) BGP Table 1.), AS1, BGP Table 1.), AS2, AS1, (AS2, AS1), BGP Table 1.), AS1, BGP Routing, v BGP Update (5) Adj-RIBs-In, Local-RIB Adj-RIBs-In, Local-RIB BGP Table 1.), AS1, 2.), AS2, AS1, (I) (AS2, AS1), BGP Table 2.), AS2, AS1, 1.), AS1, (I) new path longer than old path must not be propagated to but advertisement still stored as second choice in RIB (AS1), new path shorter than old path must be propagated to because BGP table has changed; old advertisement still stored as second choice in RIB BGP Routing, v Page 47-11 IGP Routing Update AS3 (6) IGP Routing Update BGP Routing, v BGP Update Meshed (7) BGP Table 2.), AS2, AS1, 1.), AS1, (I) (AS3, AS1), BGP Table 1.), AS1, (I) 2.), AS3, AS1, BGP Routing, v Page 47-12 BGP Final Picture (8) BGP BGP Table Table 1.), AS1, 2.), AS2, AS1, (I) 1.) means best BGP route, which is stored in the Local-RIB 2.) means alternate BGP route which is stored in Adj-RIBs-In BGP Table, internal BGP Table 2.), AS2, AS1, 1.), AS1, (I) BGP Table 1.), AS1, (I) 2.), AS3, AS1, BGP Table 1.), AS1, BGP Routing, v Agenda BGP Internals BGP Session Topologies Synchronization with IGP Fail-over Handling BGP Routing, v Page 47-13 BGP Start Picture (Scenario 1) BGP BGP Table Table 1.), AS1, 2.), AS2, AS1, (I) BGP Table 2.), AS2, AS1, 1.), AS1, (I) BGP Table, internal BGP Table 1.), AS1, (I) 2.), AS3, AS1, BGP Table 1.), AS1, BGP Routing, v Session Loss (Scenario 1) BGP Table, AS1, 1.), AS2, AS1, (I) information removed from table; use second choice BGP Table 1.), AS2, AS1,, AS1, (I) new path from currently selected neighbor use second choice withdraw BGP Routing, v Page 47-14 IGP Routing Update AS3 (1) BGP Table 1.), AS2, AS1,, AS1, (I) table was modified and update must be sent to corresponding peer BGP Routing, v BGP Incremental Update (2) BGP Table 1.), AS1, (I), AS3, AS1, withdraw BGP Routing, v Page 47-15 BGP Final Picture (3) BGP BGP Table Table 1.), net AS2, 10, AS1, AS1, (I) BGP Table 1.), AS2, AS1, BGP Table, internal BGP Table 1.), AS1, (I) BGP Table 1.), AS1, BGP Routing, v BGP Start Picture (Scenario 2) BGP BGP Table Table 1.), AS1, 2.), AS2, AS1, (I) BGP Table 2.), AS2, AS1, 1.), AS1, (I) BGP Table, internal BGP Table 1.), AS1, (I) 2.), AS3, AS1, BGP Table 1.), AS1, BGP Routing, v Page 47-16 Session Loss (Scenario 2) BGP BGP Table Table 1.), AS1, 2.), AS2, AS1, (I) BGP Table 2.), AS2, AS1, 1.), AS1, (I) BGP Table, internal BGP Table 1.), AS1, (I) 2.), AS3, AS1, BGP Routing, v BGP Table 1.), AS1, withdraw BGP Incremental Update (1) BGP BGP Table Table 1.), AS1, 2.), AS2, AS1, (I) BGP Table 2.), AS2, AS1, 1.), AS1, (I) BGP Table, internal BGP Table 1.), AS3, AS1, BGP Routing, v BGP Table 1.), AS3, AS1, (I), Page 47-17 IGP Routing Update (2) BGP BGP Table Table 1.), AS1, 2.), AS2, AS1, (I) BGP Table 2.), AS2, AS1, 1.), AS1, (I) BGP Table, internal BGP Table 1.), AS3, AS1, BGP Table 1.), AS3, AS1, (I) BGP Routing, v BGP Incremental Update (3) BGP BGP Table Table 1.), AS1, 2.), AS2, AS1, (I) withdraw BGP Table 2.), AS2, AS1, 1.), AS1, (I) withdraw BGP Table, internal BGP Table 1.), AS3, AS1, BGP Table 1.), AS3, AS1, (I) BGP Routing, v Page 47-18 BGP Final Picture (4) BGP BGP Table Table 1.), 10, AS1, BGP Table 1.), AS1, (I) BGP Table, internal BGP Table 1.), AS3, AS1, BGP Routing, v BGP Table 1.), AS3, AS1, (I) Page 47-19
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks