Shulou(Shulou.com)06/02 Report--

External gateway protocol, TCP as transport layer protocol, CIDR support, incremental update, distance vector routing protocol, no loop, rich routing policy, can prevent routing shock, easy to expand.

BGP Overview

#Message types of how BGP works TCP three-way handshake open message (type "1", negotiate BGP parameters, including version, AS number, hold time, etc., and then establish neighbors) update message (type "2", exchanging routing information) contains NLRI network layer reachability information keepalive message (type "4", keep neighbor relationship) Notification message (type "3", error notification) Route-Refresh message (route refresh message: used to request peers to resend routing information after changing routing policy)(command: refresh bgp all import)#BGP working database IP routing table BGP routing table Neighbor table (peer list) adj-RIB-In (unprocessed routing information base announced by peers to this AS) adj-RIB-Out (the routing information base announced by the local speaker to the specified peer)

How BGP Works-BGP Routing Information Processing

BGP considerations

1: Due to the horizontal segmentation of BGP, there will be routing black holes in BGP. There are three solutions (IBGP full interconnection, RR, federation). 2: EBGP neighbor, default TTL = 1 3: The IGP routing table is checked before the IBGP route is added to the routing table and published to the EBGP peer. Only if the IGP also knows about the IBGP route will it be added to the routing table and published to the EBGP peer. Synchronization: IGP table and BGP table synchronization, will be sent to EBGP peer 4: Private AS number (64512-65535) 5: When static routes are introduced using the import-route static command, only active routes can be introduced.

BGP attribute classification and enumeration explanation

#1: Recognized compliance (as-path, next_hop, origin) #2: Recognized arbitrary (Local_Pref local priority affects traffic, atomic aggregation) #Third: Optional transfer (aggregator, community attributes similar to tag) #Fourth: Optional non-transitive (MED multi-exit discriminator affects inbound traffic of local AS, Cluster List, Originator ID)#A: origin attribute (used to define the source of path information) IGP(i): Routes introduced through the network will appear as i in the bgp routing-table. EGP(e): Route obtained through EGP, will be displayed as e (no longer common) Incomplete(?)：Route information learned by other means is represented as? The principle of selection: I>E>? #Three Ways to Join BGP import and route aggregation network: Directly connected routes can be added in such a way (i.e., when the route we need to join BGP exists in the IGP routing table, this can be used, directly connected) #B: as-path attribute (used for routing, used to define which paths are passed, can be used for ebgp anti-loop) #aspath There are 4 types of as records displayed by packet capture. as-seq Record the as number (ordinary) in a sequential manner as-set is used to prevent loops when BGP aggregates The other two are used in the alliance #C: next_hop next hop attribute: (with pictures) 1: The routing information learned from IBGP will not be passed to other IBGP peers by default (normal, full interconnection model can be established) 2: Route learned from IBGP peer, not changed in next hop when passed to EBGP peer (normal) 3: When the route learned from EBGP is delivered to the IBGP peer, the next hop does not change by default (abnormal, which will cause the route of the IBGP peer to be unreachable in the next hop). The solution is #D: The Local_Pref attribute indicates the priority of BGP routing information, which is used to determine the best route when traffic leaves the AS (the larger the value, the higher the priority, and the default is 100). 1: Traffic that can only be passed between IBGP and affects this AS #E: community attribute (written 100:1)(used to identify BGP routes with the same characteristics) 1: Custom groups (usually passed internally) 2: Public group attributes Internet Internet properties, as long as it is a public route, all have No Advertise Do not advertise routes (if route information with this value is received, the route will not be passed on to any other BGP neighbors) No Export No routing (if routing information with this value is received, it will not be passed to any EBGP neighbors) No Export Subconfessed is used for federation, and the route is not passed to federation EBGP (i.e., not passed to normal EBGP neighbors)#F: MED multi-exit discriminator, equivalent to link cost 1:(Huawei default is 0) affects inbound traffic (note that I am talking about traffic, not routing), used for EBGP neighbors (the smaller the value of Huawei equipment, the more priority), can be transmitted within the peer AS, and the peer AS cannot be transmitted to other AS 2:(Reminder: affecting inbound traffic means affecting the other party, that is to say, the AS of the other party can be compared)#Note that there are origin attributes, AS path, local_ref, MED that affect routing. Others such as commity simply control without directly affecting routing

BGP Extension Features

BGP Thirteen Routing Principles

#When there are multiple routes to the same destination, BGP routes according to the following policy order 1: Ignore this route if its next hop is unreachable 2: Preferred Protocol Preferred Value (PrefVal) Highest Route (Huawei Private Feature) 3: Prefer the route with the highest Local_Pref 4: Locally generated routing is preferred (i.e., locally originated, locally aggregated) 5: Select the shortest route for AS 6: Compare Origin attributes, where i>e>? network>ebgp>import 7: Choose the route with the lowest MED 8: Prefer routes learned from EBGP neighbors (EBGP routes take precedence over IBGP routes) 9: Preferred route to next hop IGP Metric smaller 10: Select the shortest route in Cluster List (Load Balancer is possible only if all the previous 9 are the same) 11: Choose the route published by the router with the smallest Router ID 12: Compare IP addresses of peers, prefer routes learned from peers with smaller IP addresses BGP how to achieve Load Balancer, the first 8 items are not compared to determine whether priority or not MED comparisons can only occur within the same AS

BGP Advanced Feature Description

#Route reflector (refers to IBGP, simple implementation, does not change the existing network topology) 1: Root cause: Horizontal split principle, transit AS, routing black hole 2: Role type: RR route reflector, client client, non-client 3: Reflection rules As long as there is a route on RR, it can pass it to all clients. 4: route reflector-anti-ring mechanism (cluster list) A: Originator ID attribute (optional non-transitive), used for ring prevention in the cluster, generated by route reflector RR, carrying the Router ID of the route sender in the local AS. B: Cluster List attribute (optional non-transitive), used for anti-ring between clusters, with each route reflector RR generated to record the cluster through which the reflected route passes. 5: Route Reflector-Backup RR Route reflectors in the same cluster share the same Cluster ID The application of Cluster List ensures that routing loops do not occur between different RRs within the same AS. 6: Route Reflector-Peer RR There may be multiple clusters in an AS, and there are IBGP peer relationships between RRs. 7: Route Reflector-Graded RR Configure RR at lower network levels to client#BGP federation (less used) for RR at higher network levels 1. Divide an AS into several sub-ASs, establish a fully interconnected BGP neighbor relationship within each sub-AS, and establish EBG connection relationships among sub-ASs. 2: Alliance-Anti-ring mechanism (record type of AS_Path) AS_CONFED_SEQUENCE AS_CONFED_SET#Route aggregation (automatic aggregation: only imported routes can be aggregated, not network routes) Send aggregated routes only to peers, reducing routing table size If the detailed route fluctuates, it will not affect the network. Route aggregation is divided into automatic aggregation and manual aggregation https://article.pchome. net/content-2099609.htmlhttps://www.csdn. net/article/a/2019-09-02/15980449#BGP enhancement features 1: BGP ORF(Outbound Route Filtering): outbound route filtering (with figure) BGP is based on prefix ORF and local ingress policy to construct egress policy at the opposite end, so as to realize BGP on-demand published route. Including prefix-based ORFs and ORFs 2：Active-Route-Advertise BGP routes can only be sent to neighbors if they are successfully installed into the IP routing table 3: BGP packaged by group The group-by-group packaging technique treats all BGP neighbors with a common egress policy as a packaged group Each route to be sent is packaged only once and sent to all neighbors in the group. 4:4 byte AS number 5: Next hop iteration by strategy 1: Limit the routes iterated to by configuring routing policies. If the route fails to pass the routing policy, the route iteration fails.

Internet Design Principles

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.