Cours BGP-MPLS-IPV6-QOS France Telecom Introduction to BGP

Outline Reviewing Routing Operations (1-3) Interdomain Routing (1-9) BGP Characteristics (1-13) Single-Homed and Multihomed Customers (1-18) Selecting a BGP Path (1-25) Working with a Transit AS (1-39) Interacting with IBGP and EBGP in a Transit AS (1-46) Processing BGP Routes (1-61)

Reviewing Routing Operations

Static vs. Dynamic Routes Static Route Uses a route that a network administrator enters into the router manually Dynamic Route Uses a route that a network routing protocol adjusts automatically for topology or traffic changes

What Is a Dynamic Routing Protocol? Routing protocols are used between routers to determine paths to remote networks and maintain those networks in the routing tables. After the path is determined, a router can route a routed protocol to the learned networks

Autonomous Systems: IGP and EGP An autonomous system is a collection of networks within a common administrative domain. Interior gateway protocols operate within an autonomous system. Exterior gateway protocols connect different autonomous systems.

Classes of Routing Protocols

Selecting the Best Route Using Metrics

Interdomain routing

Interdomain Routing An AS is a collection of networks under a single technical administration. An IGP is run inside an AS, resulting in optimum intra-AS routing. An EGP is run between autonomous systems to enable routing policies and improve security.

Design Goals for Interdomain Routing Scalability The Internet has more than 300,000 routes and is still growing. Secure routing information exchange Routers from another AS cannot be trusted. Tight filters are required; authentication is desirable. Support for routing policies Routing between autonomous systems might not always follow the optimum path.

Why External Routing Protocols? Q: Assuming standard IGP route selection rules, how will the traffic between AS 1 and AS 20 flow? Q: Will AS 2 allow this traffic? Q: How would you solve this problem with OSPF or EIGRP?

BGP characteristics

BGP Characteristics BGP is a distance vector protocol with enhancements: Reliable updates Triggered updates only Rich metrics (1-called path attributes) Designed to scale to huge internetworks

BGP Characteristics (1-Cont.) Reliable updates TCP 179 used as transport protocol No periodic updates Periodic keepalives to verify TCP connectivity Triggered updates batched and rate-limited Every 5 seconds for internal peer Every 30 seconds for external peer

BGP Characteristics (1-Cont.) Protocol development considerations BGP was designed to perform well in the following areas: Interdomain routing applications Huge internetworks with large routing tables Environments that require complex routing policies Some design tradeoffs were made: BGP uses TCP for reliable transport— CPU-intensive Scalability is the top priority—slower convergence

BGP Characteristics (1-Cont.) Common BGP uses Customers connected to more than one service provider Service provider networks (1-transit autonomous systems) Service providers exchanging traffic at an exchange point (1-CIX, GIX, NAP, …) Network cores of large-enterprise customers

Single-Homed and Multi-Homed Customers

Single-Homed Customers Large customer or small ISP connecting to the Internet

Use Guidelines―Single-Homed Customers Use BGP between the customer and the service provider in these situations: Customers multihomed to the same service provider Customers that need dynamic routing protocol with the service provider to detect failures Hint: Use private AS number for these customers. Smaller ISPs that need to originate their routes in the Internet Use static routes in all other cases: Static routes always simpler than BGP

Multihomed Customers Customer connecting to more than one service provider

User Guidelines―Multihomed Customers BGP is almost mandatory for multihomed customers. Multihomed customers have to use public AS numbers. Multihomed customers should use a provider-independent address space.

Transit Autonomous Systems Using BGP to exchange routes is mandatory for transit autonomous systems (1-provider networks carrying customer traffic).

BGP Limitations BGP and associated tools cannot express all routing policies. You cannot influence the routing policies of downstream autonomous systems. “BGP does not enable one AS to send traffic to a neighbor AS intending that the traffic take a different route from that taken by traffic originating in the neighbor AS.” RFC 1771

Selecting a BGP Path

BGP Path Attributes BGP metrics are called path attributes. Characteristics of path attributes include: Well-known versus optional Mandatory versus discretionary Transitive versus nontransitive Partial

Well-Known Attributes Must be recognized by all compliant BGP implementations Are propagated to other neighbors Well-known mandatory attributes Must be present in all update messages Well-known discretionary attributes May be present in update messages

Optional Attributes Optional attributes They are recognized by some implementations (1-could be private); but expected not to be recognized by all BGP routers. Recognized optional attributes are propagated to other neighbors based on their meaning. Optional transitive attributes If not recognized, marked as partial and propagated to other neighbors Optional nontransitive attributes Discarded if not recognized

BGP Attributes BGP attributes include the following: AS path * Next-hop * Origin * Local preference MED Others * Well-known mandatory attribute

AS Path Attribute A list of autonomous systems that a route has traversed: For example, on router B, the path to 192.168.1.0 is the AS sequence (1-65500, 64520). The AS path attribute is well-known, mandatory.

Next-Hop Attribute The IP address of the next AS to reach a given network: Router A advertises network 172.16.0.0 to router B in EBGP, with a next hop of 10.10.10.3. Router B advertises 172.16.0.0 in IBGP to router C, keeping 10.10.10.3 as the next-hop address. The next-hop attribute is well-known, mandatory.

Origin Attribute IGP (1-i) network command EGP (1-e) Redistributed from EGP Incomplete (1-?) Redistributed from IGP or static The origin attribute informs all autonomous systems in the internetwork how the prefixes were introduced into BGP. The origin attribute is well-known, mandatory.

Example: Origin Attribute RouterA# show ip bgp BGP table version is 14, local router ID is 172.31.11.1 Status codes: s suppressed, d damped, h history, * valid, > best, i - internal, r RIB-failure, S Stale Origin codes: i - IGP, e - EGP, ? - incomplete Network Next Hop Metric LocPrf Weight Path *> 10.1.0.0/24 0.0.0.0 0 32768 i * i 10.1.0.2 0 100 0 i *> 10.1.1.0/24 0.0.0.0 0 32768 i *>i10.1.2.0/24 10.1.0.2 0 100 0 i *> 10.97.97.0/24 172.31.1.3 0 64998 64997 i * 172.31.11.4 0 64999 64997 i * i 172.31.11.4 0 100 0 64999 64997 i *> 10.254.0.0/24 172.31.1.3 0 0 64998 i * 172.31.11.4 0 64999 64998 i * i 172.31.1.3 0 100 0 64998 i r> 172.31.1.0/24 172.31.1.3 0 0 64998 i r 172.31.11.4 0 64999 64998 i r i 172.31.1.3 0 100 0 64998 i *> 172.31.2.0/24 172.31.1.3 0 0 64998 i <output omitted>

Local Preference Attribute Paths with highest local preference value are preferred: Local preference is used to advertise to IBGP neighbors about how to leave their AS. The local preference is sent to IBGP neighbors only (1-that is, within the AS only). The local preference attribute is well-known and discretionary. Default value is 100.

MED Attribute The paths with the lowest MED (1-also called the metric) value are the most desirable: MED is used to advertise to EBGP neighbors how to exit their AS to reach networks owned by this AS. The MED attribute is optional and nontransitive.

Weight Attribute (1-Cisco Only) Paths with the highest weight value are preferred Weight not sent to any BGP neighbors; local to this router only

BGP Path Selection The BGP forwarding table usually has multiple paths from which to choose for each network. BGP is not designed to perform load balancing: Paths are chosen because of policy. Paths are not chosen based on bandwidth. The BGP selection process eliminates any multiple paths through attrition until a single best path is left. That best path is submitted to the routing table manager process and evaluated against the methods of other routing protocols for reaching that network (1-using administrative distance). The route from the source with the lowest administrative distance is installed in the routing table.

Route Selection Decision Process Consider only (1-synchronized) routes with no AS loops and a valid next hop, and then: Prefer highest weight (1-local to router). Prefer highest local preference (1-global within AS). Prefer route originated by the local router (1-next hop = 0.0.0.0). Prefer shortest AS path. Prefer lowest origin code (1-IGP < EGP < incomplete). Prefer lowest MED (1-exchanged between autonomous systems). Prefer EBGP path over IBGP path. Prefer the path through the closest IGP neighbor. Prefer oldest route for EBGP paths. Prefer the path with the lowest neighbor BGP router ID. Prefer the path with the lowest neighbor IP address.

Working with a transit AS

Working with a Transit AS Overview Transit AS Tasks External Route Propagation Internal Route Propagation Packet Forwarding in an AS Core Router IBGP Requirements in a Transit AS

Transit AS Tasks Propagate routes between remote autonomous systems Route packets between remote networks

External Route Propagation

Internal Route Propagation

Packet Forwarding in an AS

Core Router IBGP Requirements in a Transit AS All core routers must have all external routes. Core routers must receive BGP routes. Redistribution of BGP routes into IGP is not scalable. Default routing is not applicable in transit AS core.

BGP Transit Autonomous Systems Interacting with IBGP and EBGP in a Transit AS

Outline Overview AS-Path Processing in IBGP Multipath Load Sharing in BGP BGP Split Horizon IBGP Full Mesh IBGP Neighbors IBGP Next-Hop Processing Transit Network Using External Next Hops Transit Network Using Edge Routers as Next Hops Differences Between EBGP and IBGP Sessions Summary

AS-Path Processing in IBGP

Multipath Load Sharing in BGP

BGP Split Horizon Result: Full mesh of IBGP sessions is required for proper IBGP update propagation.

IBGP Full Mesh Full mesh of IBGP sessions has to be established between all BGP-speaking routers in the AS for proper IBGP route propagation. The IBGP full mesh is a logical mesh of TCP sessions only; physical full mesh is not required.

IBGP Full Mesh Example

IBGP Neighbors Because of IBGP full-mesh requirements, IBGP neighbors are usually not directly connected. Which interfaces should be chosen as the source and destination addresses of IBGP TCP sessions?

IBGP Neighbors (1-Cont.) Always run IBGP sessions between loopback interfaces. IBGP sessions can always be established, even if some physical interfaces are down. IBGP sessions are stable—physical interface failure will not tear down IBGP sessions. There is no BGP recovery after a failure inside the transit AS. The configured IGP will re-establish the path between loopback interfaces. IBGP sessions are not affected.

IBGP Next-Hop Processing

Transit Network Using External Next Hops All EBGP peers must be reachable by all BGP-speaking routers within the AS. EBGP next hops shall be announced using the IGP. Redistribute connected interfaces into the IGP at the edge routers. or Include links to EBGP neighbors into the IGP and configure them as passive interfaces.

Transit Network Using Edge Routers as Next Hops Alternate design: Next-hop processing is modified at the edge routers. Edge routers announce themselves as the next hop in IBGP updates. No redistribution of external subnets is necessary. This design might result in suboptimal routing if multiple paths to a neighboring AS exist. Use default next-hop processing if at all possible.

Transit Network Using Edge Routers as Next Hops Example

Differences Between EBGP and IBGP Sessions No BGP attributes are changed in IBGP updates. Because of BGP split horizon, routes learned from an IBGP peer are not advertised to other IBGP peers. Local preference is propagated only over IBGP sessions. EBGP peers are directly connected; IBGP peers are usually distant. Route selection rules slightly prefer EBGP routes.

Differences Between EBGP and IBGP Sessions Example Whenever identical routes are received from IBGP and EBGP peers, the route from the EBGP peer is preferred.

BGP Practical Processing BGP Routes

Outline Overview Receiving Routing Updates Building the BGP Table BGP Route Selection Criteria BGP Route Propagation Building the IP Routing Table Advertising Local Networks Automatic Summarization Summary

Receiving Routing Updates Small BGP Network

Receiving Routing Updates (1-Cont.) Information from the BGP tables is exchanged after adjacency establishment.

Building the BGP Table All inbound updates are placed into the BGP table.

BGP Route Selection Criteria Exclude routes with inaccessible next hop Prefer highest weight (1-local to router) Prefer highest local preference (1-global within AS) Prefer routes that the router originated Prefer shortest AS path (1-only length is compared) Prefer lowest origin code (1-IGP < EGP < Incomplete) Prefer lowest MED Prefer external (1-EBGP) paths over internal (1-IBGP) For IBGP paths, prefer path through closest IGP neighbor For EBGP paths, prefer oldest (1-most stable) path Prefer paths from router with the lowest BGP router-ID

BGP Route Selection Criteria (1-Cont.) The best routes to the destination networks are selected from the BGP table.

BGP Route Propagation The best BGP routes are propagated to BGP neighbors.

Building the IP Routing Table The best BGP routes are copied into the IP routing table based on administrative distance.

Advertising Local Networks The BGP router process keeps a list of local networks (1-defined with the network command or through redistribution). The BGP process periodically scans the IP routing table and inserts or revokes routes from the BGP routing table based on their presence in the IP routing table.

Advertising Local Networks (1-Cont.) The BGP route is revoked after the network is removed from the routing table.

Advertising Local Networks (1-Cont.) The BGP route is advertised after the network appears in the routing table.