BGP - 2 - BGP Routing Policies

Route Filtering and Route Summarization

Four popular tools used to filter BGP routes:

1. Distribution lists
2. Prefix lists
3. AS_PATH filter lists
4. Route maps

Additionally, the aggregate-address.

The four main tools have the following features in common:

• All can filter incoming and outgoing Updates, per neighbor or per peer group.
• Peer group configurations require Cisco IOS Software to process the routing policy against the Update only once, rather than once per neighbor.
• The filters cannot be applied to a single neighbor that is configured as part of a peer group.
• Each tool’s matching logic examines the contents of the BGP Update message, which includes the BGP PAs and network layer reachability information (NLRI).
• If a filter’s configuration is changed, a clear command is required for the changed filter to take effect.
• The clear command can use the soft reconfiguration option.

Filtering BGP Updates Based on NLRI

One difference between BGP distribute lists and IGP distribute lists is that a BGP distribute list can use an extended ACL to match against both the prefix and the prefix length. When used with IGP filtering tools, ACLs called from distribute lists cannot match against the prefix length.

The prefix list matches the exact prefixes and prefix lengths; the omission of any ge or le parameter means each line matches only that exact prefix.

Both the route map and any referenced ACL or prefix list have deny and permit actions configured. The route-map command’s action—either deny or permit —defines whether an NLRI is filtered ( deny ) or allowed to pass ( permit ). The permit or deny action in an ACL or prefix list implies

whether an NLRI matches the route map clause ( permit by the ACL/prefix list) or does not match ( deny in the ACL/prefix list).

To support soft reconfiguration, BGP must remember the actual sent and received BGP Update information for each neighbor. The neighbor neighbor-id soft-reconfiguration inbound command causes the router to keep a copy of the received Updates from the specified neighbor. (IOS keeps a copy of sent Updates automatically.) 

For configuration changes that impact the local injection of routes into the BGP table, soft reconfiguration does not help. The reason is that soft

reconfiguration simply reprocesses Updates, and features that inject routes into BGP through the redistribute or network commands are not injected based on Update messages.

Comparing BGP Prefix Lists, Distribute Lists, and Route Maps

If the desired policy is only to filter routes based on matching prefixes/lengths, a route map does not provide any additional function over using a distribute list or prefix list directly. Similarly, if the goal of the policy is to filter routes just based on matching with an AS_PATH filter, the route map does not provide any additional function as compared to calling an AS_PATH filter directly using the neighbor filter-list command. However, only route maps can provide the following two functions for BGP routing policy configurations:

• Matching logic that combines multiples of the following: prefix/length, AS_PATH, or other BGP PAs.
• The setting of BGP PAs for the purpose of manipulating BGP’s choice of which route to use

Filtering Subnets of a Summary Using the aggregate-address Command

The filtering options on the aggregate-address command are as follows:

• Filtering all component subnets of the summary from being advertised, by using the summary-only keyword
• Advertising all the component subnets of the summary, by omitting the summaryonly keyword
• Advertising some and filtering other component subnets of the summary, by omitting the summary-only keyword and referring to a route map using the suppressmap keyword.

Filtering BGP Updates by Matching the AS_PATH PA

To filter routes by matching the AS_PATH PA, Cisco IOS uses AS_PATH filters.

The main two steps are as follows:

1. Configure the AS_PATH filter using the ip as-path access-list number { permit | deny } regex command.
2. Enable the AS_PATH filter using the neighbor neighbor-id filter-list as-path-filter-number { in | out } command.

Because the most recently added ASN is the first ASN in the AS_SEQUENCE segment, the process of adding the ASN before advertising routes to external BGP (eBGP) peers is called AS_PATH prepending. 

Including the as-set keyword, R4 creates an AS_SET segment in the AS_PATH of the aggregate route. Note that the AS_SET segment is shown in brackets, and it is listed in no particular order. These facts are all important to the process of AS_PATH filtering.

Confederation ASNs are used to prevent loops inside the confederation. Because these ASNs will be removed before advertising the route outside the full AS, the confederation ASNs are kept inside a different segment—the AS_CONFED_SEQ segment. Finally, if a route is aggregated inside a confederation, the AS_CONFED_SET segment holds the confederation ASNs with the same logic as used by the AS_SET segment type, but keeps them separate for easy removal before advertising the routes outside the confederation.

The show ip as-path-access-list command shows the contents of the list.

The show ip bgp neighbor neighbor-id advertised-routes command displays the routes actually sent—in other words, this command reflects the effects of the filtering by omitting the filtered routes from the output.

The show ip bgp neighbor neighbor-id received-routes command displays the routes actually received from a neighbor, never omitting routes from the output, even if the router locally filters the routes on input.

Output filter lists are applied before the router adds its own ASN to the AS_PATH.

A couple of ways to test regex without changing the routing policy.

show ip bgp neighbor 10.1.34.4 received-routes | include 4_1_.*_.*_.*_44

This command parses the entire command output using the regex after the include keyword.

The other method to test a regex is to use the show ip bgp regexp expression command. This command parses the AS_PATH variables in a router’s BGP table, including all special characters. However, the regexp option of the show ip bgp command is not allowed with the received-routes or advertised- routes option.

Note that the "(" must be matched by enclosing it in square brackets, as ! the "(" itself and the ")" are metacharacters, and would otherwise be interpreted as a metacharacter. Without the "[(]" to begin the regex, the ! AS_PATH filter would not match. Because the "{" and "}" are not metacharacters, they can simply be typed directly into the regex.

BGP Path Attributes and the BGP Decision Process

Each BGP PA can be described as either a well-known or optional PA. 

Well-known PAs are either one of the following:

• Mandatory: The PA must be in every BGP Update.
•  Discretionary: The PA is not required in every BGP Update.

The BGP Decision Process

1. Is the NEXT_HOP reachable?
2. Highest administrative weight
3. Highest LOCAL_PREF PA
4. Locally injected routes
5. Shortest AS_PATH length: The length calculation ignores both AS_CONFED_SET and AS_CONFED_SEQ, and treats an AS_SET as one ASN, regardless of the number of ASNs in the AS_SET. It counts each ASN in the AS_SEQUENCE as one. (This step is ignored if the bgp bestpath as-path ignore command is configured.)
6. ORIGIN PA
7. Smallest Multi-Exit Discriminator (MED) PA: The smaller the value, the better the route.
8. Neighbor Type: Prefer external BGP (eBGP) routes over internal BGP (iBGP).
9. IGP metric for reaching the NEXT_HOP.

If a step determines the best route for an NLRI, BGP does not bother with the remaining steps.

When overlapping NLRIs exist—for example, 130.1.0.0/16, 130.2.0.0/16, and 130.0.0.0/12—BGP attempts to find the best route for each specific prefix/prefix length.
First and last of the nine items relate to NEXT_HOP.

Configuring BGP Policies

The show ip bgp <network> command lists the advertising router’s RID and neighbor ID.

The "from z.z.z.z" phrases identify the neighbor ID that advertised the route. The "(y.y.y.y)" output that follows lists the RID of that same router.

Step 1: NEXT_HOP Reachable

Step 2: Administrative Weight

Default 0 for learned routes, 32,768 for locally injected routes

The neighbor route-map command creates an implied filtering decision. Any route matched by a permit clause in the route map is implied to be allowed through, and routes matched by a deny clause will be filtered. Route maps use an implied deny all at the end of the route map for any unmatched routes. By including a final clause with just a permit keyword, the route map changes to use permit all logic, thereby passing all routes.

Step 3: Highest Local Preference (LOCAL_PREF)

Changing the default Using the bgp default local-preference <0-4294967295> BGP subcommand

Step 4: Choose Between Locally Injected Routes Based on ORIGIN PA

When the same NLRI is locally injected into BGP from multiple methods, pick the route with the better ORIGIN PA.

Step 5: Shortest AS_PATH

bgp bestpath as-path ignore command - Removes the AS_PATH length step from the decision tree for the local router.

Removing Private ASNs - 

 Private ASNs can be removed only at the point of sending an eBGP Update.

 If the current AS_SEQ contains both private and public ASNs, the private ASNs will not be removed.

 If the ASN of the eBGP peer is in the current AS_PATH, the private ASNs will not be removed, either.

The aggregate-address command with the as-set option can lengthen the AS_PATH length calculation as well.

The BGP AS_PATH length calculation counts the entire AS_SET as 1, regardless of the actual length.

Step 6: Best ORIGIN PA

The well-known mandatory BGP ORIGIN PA characterizes a route based on how it was injected into BGP. 

If the set of routes to reach a single NLRI includes only one route of ORIGIN code IGP (i), and all the others as incomplete (?), the route with ORIGIN i is the best route. BGP routing policies can set the ORIGIN code explicitly by using the set origin route

map subcommand, although the earlier steps in the BGP decision process are typically better choices.

Step 7: Smallest Multi-Exit Discriminator

Scope - Advertised by one AS into another, propagated inside the AS but not sent to any other ASs. Smaller is better.

The purpose of the MED (or MULTI_EXIT_DISC) is to allow routers in one AS to tell routers in a neighboring AS how good a particular route is. A default setting is 0. MED. A better default for MED can be set by using the bgp bestpath med missing-as-worst BGP subcommand, which resets a router’s default MED to the largest possible MED value, instead of the lowest.

Configuring MED: Multiple Adjacent Autonomous Systems

By default, a Cisco router ignores MED when the multiple routes to a single NLRI list different neighboring ASNs. This default action makes sense—normally you would not expect two different neighboring ISPs to have chosen to work together to set MEDs. To override this default and consider the MED in all cases, a router needs to configure the bgp always-compare-med BGP subcommand. If used on one router, all routers inside the

same AS should also use the bgp always-compare-med command, or routing loops can result.

After reaching the other AS, the MED is advertised inside the AS, but not outside the AS.

MED can also be set through inbound route maps, although that is not the intended design with which to use MED.

Step 8: Prefer Neighbor Type eBGP over iBGP

BGP uses this decision point frequently when two or more enterprise routers connect to the same ISP.

Each enterprise border router knows of one eBGP route to reach each prefix, and one or more iBGP routes to the same prefix learned from that enterprise’s other border routers. With no routing policies configured, the routes tie on all decision points up to this one, including AS_PATH length, because all the prefixes were learned from the same neighboring ISP. The decision process reaches this step, at which point the one eBGP route is picked as the best route.

Step 9: Smallest IGP Metric to the NEXT_HOP

Step 10: Lowest BGP Router ID of Advertising Router

Step 11: Lowest Neighbor ID

The BGP maximum-paths Command

BGP defaults the maximum-paths command to a setting of 1. However, BGP will consider adding multiple entries to the IP routing table, for the same NLRI, under certain conditions—conditions that differ based on whether the best route is an eBGP route or an iBGP route.

The following rules determine if and when a router will add multiple eBGP routes to the IP routing table for a single NLRI:

1. BGP must have had to use a tiebreaker (Step 10 or 11) to determine the best route.
2. The maximum-paths number command must be configured to something larger than the default of 1.
3. Only eBGP routes whose adjacent ASNs are the same ASN as the best route are considered as candidates.
4. If more candidates exist than that called for with the maximum-paths command, the tiebreakers of Steps 10 and 11 determine the ones to use.

The rules for iBGP have some similarities with eBGP, and a few differences, as follows:

1. Same rule as eBGP rule 1.
2. The maximum-paths ibgp number command defines the number of possible IP routes, instead of the maximum-paths number command used for eBGP.
3. Only iBGP routes with differing NEXT_HOP settings are considered as candidates.
4. Same rule as eBGP rule 4.

BGP Communities

The BGP COMMUNITY PA provides a mechanism by which to group routes so that routing policies can be applied to all the routes with the same community. 

BGP communities are powerful in that they allow routers in one AS to communicate policy information to routers that are one or more autonomous systems distant. In fact, because the COMMUNITY PA is an optional transitive PA, it can pass through autonomous systems that do not even understand the COMMUNITY PA and then still be useful at another downstream AS.

The only way to match the COMMUNITY is to refer to an ip community-list , which then has the matching parameters.

The set community 10 20 30 additive command would add the values to the existing COMMUNITY string.

The show ip bgp community-list list-number command is then used to show whether a match would be made. This command lists the entries of the BGP table that match the associated COMMUNITY PA, much like the show ip bgp regex command examines the AS_PATH PA.

The set community none command in a route-map clause, and all routes matched by that clause will have their COMMUNITY PA removed. A route map can also remove individual COMMUNITY strings by using the set commlist community-list-number delete command.

Filtering NLRIs Using Special COMMUNITY Values

A route with COMMUNITY NO_EXPORT is not advertised outside an AS. This value can be used to prevent an AS from being a transit AS for a set of prefixes. 

Finally, routes with these settings can be seen with commands like show ip bgp community noexport, with similar options NO_ADVERT and LOCAL_AS.

Fast Convergence Enhancements

BGP only provides updates to its neighbors periodically using an interval based on the peering type: iBGP peers receive updates every 5 seconds, whereas eBGP peers are updated only every 30 seconds. BGP will only verify next-hop reachability every 60 seconds.

Fast External Neighbor Loss Detection

The eBGP session between directly connected eBGP neighbors will be torn down the moment that the connected subnet between the peers is lost. This will result in the immediate flushing of BGP routes, and BGP will immediately begin looking at alternate routes. 

Internal Neighbor Loss Detection

With the neighbor fall-over command, the moment that the IP address of the BGP peer is removed from the routing table, the BGP session with the peer will be torn down, thus resulting in immediate convergence. 

EBGP Fast Session Deactivation

Use it to quickly detect failures of eBGP sessions established between loopback interfaces of eBGP peers or to detect eBGP neighbor loss when you disable fast external fall-over.

Summary

network ip-address backdoor 

- BGP mode; identifies a network as a backdoor route, considering it to have the same administrative distance as iBGP routes

BGP - 1 - Foundation

BGP does not use a metric to select the best route among alternate routes to the same destination. Instead, BGP uses several BGP path attributes (PA). BGP uses the BGP autonomous system path (AS_PATH) PA as its default metric mechanism when none of the other PAs has been overly set and configured.

After the TCP connection is established, BGP begins with BGP Open messages. After a pair of BGP Open messages has been exchanged, the neighbors have reached the established state, which is the stable state of two working BGP peers. At this point, BGP Update messages can be exchanged.

Peer-group allows fewer configuration commands, and improves processing efficiency by having to prepare only one set of outbound Update packets for the peer group. BGP builds one set of Update messages for the peer group, applying routing policies for the entire group—rather than one router at a time—thereby reducing some BGP processing and memory overhead.

For eBGP connections, Cisco IOS defaults the IP packet’s TTL field to a value of 1, based on the assumption that the interface IP addresses will be used for peering.

Checks Before Becoming BGP Neighbors
1. The router must receive a TCP connection request with a source address that the router finds in a BGP neighbor command.
2. A router’s ASN (on the router bgp asn command) must match the neighboring router’s reference to that ASN with its neighbor remote-as asn command. (This requirement is not true of confederation configurations.)
3. The BGP RIDs of the two routers must not be the same.
4. If configured, MD5 authentication must pass.

BGP uses a keepalive timer to define how often that router sends BGP keepalive messages, and a Hold timer to define how long a router will wait without receiving a keepalive message before resetting a neighbor connection. The Open message includes each router’s stated keepalive timer. If they do not match, each router uses the lower of the values for each of the two timers, respectively. Mismatched settings do not prevent the routers from becoming neighbors.

BGP Messages and Neighbor States

The desired state for BGP neighbors is the established state in which the routers have formed a TCP connection, and they have exchanged Open messages, with the parameter checks having passed. At this point, topology information can be exchanged using Update messages. If the IP addresses mismatch, the neighbors settle into an active state.

Building the BGP Table

The BGP topology table , also called the BGP Routing Information Base (RIB) , holds the network layer reachability information (NLRI) learned by BGP, as well as the associated PAs. Technically, BGP does not advertise routes; rather, it advertises PAs plus a set of NLRI that shares the same PA values. However, most people simply refer to NLRI as BGP prefixes or BGP routes.

The BGP network command instructs that router’s BGP process to do the following:

• Look for a route in the router’s current IP routing table that exactly matches the parameters of the network command; if the IP route exists, put the equivalent NLRI into the local BGP table.
• With this logic, connected routes, static routes, or IGP routes could be taken from the IP routing table and placed into the BGP table for later advertisement. When the router removes that route from its IP routing table, BGP then removes the NLRI from the BGP table, and notifies neighbors that the route has been withdrawn.

Impact of Auto-Summary on Redistributed Routes and the network Command

As it does with IGPs, the BGP auto-summary command causes a classful summary route to be created if any component subnet of that summary exists. However, unlike IGPs, the BGP auto-summary router subcommand causes BGP to summarize only those routes injected because of redistribution on that router. It simply looks for routes injected into the BGP because of the redistribute and network commands on that same router.

The logic differs slightly based on whether the route is injected with the redistribute command or the network command. The logic for the two commands is summarized as follows:

redistribute: If any subnets of a classful network would be redistributed, do not redistribute, but instead redistribute a route for the classful network.

network: If a network command lists a classful network number, with the classful default mask or no mask, and any subnets of the classful network exist, inject a route for the classful network.

For redistribution, the auto-summary command causes the redistribution process to inject only classful networks into the local BGP table, and no subnets. The network command, with auto-summary configured, still injects subnets based on the same logic. In addition to that logic, if a network command matches the classful network number, BGP injects the classful network, as long as at least any one subnet of that classful network exists in the IP routing table.

Manual Summaries and the AS_PATH Path Attribute

BGP manual summarization with the aggregate-address command can summarize based on any routes in the BGP table, creating a summary of any prefix length. It does not always suppress the advertisement of the component subnets, although it can be configured to do so. The aggregate route must include the AS_PATH PA, just like it is required for every other NLRI in the BGP table.

The AS_PATH PA consists of up to four different components, called segments , as follows:

• AS_SEQ (short for AS Sequence)
• AS_SET
• AS_CONFED_SEQ (short for AS Confederation Sequence)
• AS_CONFED_SET

When the component subnets of the summary route have differing AS_SEQ values, the router simply can’t create an accurate representation of AS_SEQ, so it uses a null AS_SEQ. However, this action introduces the possibility of creating routing loops. 

The AS_PATH AS_SET segment solves the problem when the summary route has a null AS_SEQ. The AS_SET segment holds an unordered list of all the ASNs in all the component subnets’ AS_SEQ segments.

"atomic-aggregate" refers to the fact that the ATOMIC_AGGREGATE PA has also been set; this PA simply states that this NLRI is a summary.

The following list summarizes the actions taken by the aggregate-address command when it creates a summary route:

• It does not create the summary if the BGP table does not currently have any routes for NLRI inside the summary.
• If all the component subnets are withdrawn from the aggregating router’s BGP table, it also then withdraws the aggregate. (In other words, the router tells its neighbors that the aggregate route is no longer valid.)
• It sets the NEXT_HOP address of the summary, as listed in the local BGP table, as 0.0.0.0.
• It sets the NEXT_HOP address of the summary route, as advertised to neighbors, to the router’s update source IP address for each neighbor, respectively.
• If the AS_SEQ of the component subnets differs in any way, it sets the AS_SEQ of the new summary route to null.
• When the as-set option has been configured, the router creates an AS_SET segment for the aggregate route, but only if the summary route’s AS_SEQ is null.
• It suppresses the advertisement of all component subnets if the summary-only keyword is used, advertises all of them if the summary-only keyword is omitted, or advertises a subset if the suppress-map option is configured.

Adding Default Routes to BGP

Default routes can be injected into BGP in one of three ways:

• By injecting the default using the network command
• By injecting the default using the redistribute command
• By injecting a default route into BGP using the neighbor neighbor-id defaultoriginate [ route-map route-map-name ] BGP subcommand

   When you inject a default route into BGP using the network command, a route to 0.0.0.0/0 must exist in the local routing table, and the network 0.0.0.0 command is required.

   Injecting a default route through redistribution requires an additional configuration command—default-information originate . The default route must first exist in the IP routing table.

   Injecting a default route into BGP by using the neighbor neighbor-id default-originate [ route-map route-map-name ] BGP subcommand does not add a default route to the local BGP table; instead, it causes the advertisement of a default to the specified neighbor. In fact, this method does not even check for the existence of a default route in the IP routing table by default, but it can.

ORIGIN Path Attribute

The ORIGIN PA provides a general descriptor as to how a particular NLRI was first injected into a router’s BGP table. Routes redistributed into BGP from an IGP actually have an ORIGIN code of incomplete.

BGP Update Message

If a router needs to advertise a set of NLRIs, and each NLRI has a different setting for at least one PA, separate Update messages will be required for each

NLRI. However, when many routes share the same PAs—typical of prefixes owned by a particular ISP, for example—multiple NLRIs are included in a single Update. This reduces router CPU load and uses less link bandwidth.

For a route to be a candidate to be considered best, the NEXT_HOP must be either

• 0.0.0.0, as the result of the route being injected on the local router.
• Reachable according to that router’s current IP routing table. In other words, the NEXT_HOP IP address must match a route in the routing table.

Note that the NEXT_HOP PA cannot be set through a route map. 

For the received-routes option to work, the router on which the command is used must have the neighbor neighbor-id soft-reconfiguration inbound BGP subcommand configured for the other neighbor.

These show ip bgp neighbor commands with the advertised-routes option list the BGP table entries that will be advertised to that neighbor. However, note that any changes to the PAs inside each entry are not shown in the command output.

Summary of Rules for Routes Advertised in BGP Updates

The following list summarizes the rules dictating which routes a BGP router sends in its update messages:

• Send only the best route listed in the BGP table.
• To iBGP neighbors, do not advertise paths learned from other iBGP neighbors.
• Do not advertise suppressed or dampened routes.
• Do not advertise routes filtered through configuration.

Adding eBGP Routes to the IP Routing Table

Cisco IOS Software uses simple logic when determining which eBGP routes to add to the IP routing table. 

• The eBGP route in the BGP table is considered to be a “best” route.
• If the same prefix has been learned through another IGP or through static routes, the AD for BGP external routes must be lower than the ADs for other routing source(s).

BGP sets the AD differently for eBGP routes, iBGP routes, and for local (locally injected) routes—with defaults of 20, 200, and 200, respectively.

The actual IP route added to the IP routing table contains the exact same prefix, prefix length, and next-hop IP address as listed in the BGP table—even if the NEXT_HOP PA is an IP address that is not in a connected network. As a result, the IP forwarding process might require a recursive route lookup.

Backdoor Routes(network backdoor) will use the local AD (default 200) for the eBGP-learned route to network. 

Adding iBGP Routes to the IP Routing Table

Cisco IOS has the same two requirements for adding iBGP routes to the IP routing table as it does for eBGP routes:

• The route must be the best BGP route.
• The route must be the best route (according to the AD) in comparison with other routing sources.

Additionally, for iBGP-learned routes, IOS considers the concept of BGP synchronization. 

The key to understanding BGP sync is to know that redistribution solves the routing

black-hole problem, and sync solves the problem of advertising a black-hole route to

another AS. 

The BGP sync logic controls that decision as follows: Do not consider an iBGP route in the BGP table as “best” unless the exact prefix was learned through an IGP and is currently in the routing table. The route must be IGP-learned not via own's static route.

Sync includes an additional odd requirement when OSPF is used as the IGP. If the OSPF RID of the router advertising the prefix is a different number than the BGP router advertising that same prefix, sync still does not allow BGP to consider the route to be the best route.

Disabling Sync and Using BGP on All Routers in an AS

A second method to overcome the black-hole issue is to simply use BGP to advertise all the BGP-learned prefixes to all routers in the AS. BGP needs the full mesh of iBGP peers inside an AS because BGP does not advertise iBGP routes (routes learned from one iBGP peer) to another iBGP peer. BGP offers two tools (confederations and route reflectors) that reduce the number of peer connections inside an AS, prevent loops, and allow all routers to learn about all prefixes.

Confederations

Peers inside the same sub-AS are considered to be confederation iBGP peers , and routers in different subautonomous systems are considered to be confederation eBGP peers. Confederation eBGP peer connections act like true eBGP peers in some respects. In a single sub-AS, the confederation iBGP peers must be fully meshed, because they act exactly like normal iBGP peers. 

Confederations prevent loops inside a confederation AS by using the AS_PATH PA. BGP routers in a confederation add the subautonomous systems into the AS_PATH as part of an AS_PATH segment called the AS_CONFED _SEQ. (The AS_PATH consists of up to four different components, called segments—AS_SEQ, AS_SET, AS_CONFED_ SEQ, and AS_CONFED_SET.

The following list summarizes the key topics regarding confederations:

• Inside a sub-AS, full mesh is required, because full iBGP rules are in effect.
• The confederation eBGP connections act like normal eBGP connections in that iBGProutes are advertised—as long as the AS_PATH implies that such an advertisement would not cause a loop.
• Confederation eBGP connections also act like normal eBGP connections regarding Time to Live (TTL), because all packets use a TTL of 1 by default. (TTL can be changed with the neighbor ebgp-multihop command.)
• Confederation eBGP connections act like iBGP connections in every other regard—for example, the NEXT_HOP is not changed by default.
• Confederation ASNs are not considered part of the length of the AS_PATH when a router chooses the best routes based on the shortest AS_PATH. 
• Confederation routers remove the confederation ASNs from the AS_PATH in Updates sent outside the confederation; therefore, other routers do not know that a confederation was used

Route Reflectors

In an iBGP design using RRs, a partial mesh of iBGP peers is defined. Some routers are configured as RR servers; these servers are allowed to learn iBGP routes from their clients and then advertise them to other iBGP peers. Note that only the RR server itself uses different logic, with clients and nonclients acting as normal iBGP peers.

One of the main motivations for using RRs is to allow sync to be disabled.
RR feature uses several tools to prevent loops, as follows:
CLUSTER_LIST: RRs add their cluster ID into a BGP PA called the CLUSTER_LIST before sending an Update. When receiving a BGP Update, RRs discard received
prefixes for which their cluster ID already appears. As with AS_PATH for confederations, this prevents RRs from looping advertisements between clusters.
ORIGINATOR_ID: This PA lists the RID of the first iBGP peer to advertise the route into the AS. If a router sees its own BGP ID as the ORIGINATOR_ID in a received route, it does not use or propagate the route.
Only advertise the best routes: RRs reflect routes only if the RR considers the route to be a “best” route in its own BGP table. This further limits the routes reflected by the RR. (It also has a positive effect compared with confederations in that an average router sees fewer, typically useless, redundant routes.)

Multiprotocol BGP 

Some of these configurations carry VPN-IPv4 routes, some only IPv4 routes, and others carry VPN-IPv4 and IPv4 routes. The type of BGP session and the specification of which routes the peering sessions will carry are controlled through the use of the address families.

Configure a BGP address family for each Virtual Routing and Forwarding (VRF) configured on the PE router and a separate address family to carry no IPv4 routes between PE routers. The initial BGP process, the portion of the configuration that cites no address family specifications, becomes the default address family. This default context becomes the “catch all” where any non-VRF-based or IPv4-specific sessions can be configured. Any prefixes learned or advertised in this default address family will be injected into the global routing table. The configuration of these BGP sessions is exactly the same as the standard BGP configuration with the exception that the session needs to be activated.

R1(config-router)# address-family vpnv4
R1(config-router)# neighbor 194.22.15.3 activate

The configuration of the VPNv4 address family also adds a further command to the BGP configuration to support the MP-BGP-specific extended community attributes. This command will be added by the IOS by default and is necessary because it instructs BGP to advertise the extended community attributes.

The default behavior is to send only the extended community attribute. If the network design requires the standard community attribute to be attached to these non-IPv4 prefixes, this behavior can be changed through the neighbor 194.22.15.3 send-community both command.

Note that MP-iBGP communicates these routes across the MP-iBGP sessions running between PE routers. To this end, the routing context must be configured under the BGP process to communicate to BGP which VRF prefixes it needs to advertise.

Summary

address-family vpnv4 

- BGP mode; allows the creation of the MP-BGP session necessary to form the VPNv4 session between PE devices

bgp client-to-client reflection

- BGP mode; on by default, tells an RR server to reflect routes learned from a client to other clients

default-information originate 

- BGP mode; required to allow a static default route to be redistributed into BGP

distance bgp external-distance internal distance local-distance 

- BGP mode; defines the administrative distance for eBGP, iBGP, and locally injected BGP routes

neighbor { ip-address | peer-group-name } default-originate [ route-map map-name ]

- BGP mode; tells the router to add a default route to the BGP Update sent to this neighbor, under the conditions set in the optional route map

show ip bgp injected-paths

- Exec mode; lists routes locally injected into BGP

Default Administrative Distance

This table shows default administrative distance value of routing protocols in use nowadays.

BGP Theory and Concepts; BGP Attributes

Administrative distance - EBGP = 20, IBGP = 200

BGP neighbor relationships
Neighbors are manually configured.
Neighbors start in IDLE state
 -1st stage: Active
 -2nd stage: Open sent
 -3rd stage: Open Confirmed
 -4th stage: Established

Hello message sent once every 60 seconds with a holddown of 180 seconds
Neighbors are capable of MD5 authentication

Rule of Synchronization
Routes learned via BGP must be validated by the interior routing table before they can be advertised to remote peers.

If Synchronization is turned on at R1, it receives eBGP routes from AS4300 and AS2300 but R1 does not install them in routing table.

After Synchronization is turned off at R1, it installs eBGP routes in its routing table. It does not advertise to its iBGP peer due to Rule of Split-Horizon.

Rule of Split-Horizon (Per BGP)
Routes learned via IBGP will never be sent to another IBGP peer. It's to prevent loop and Split-horizon is good if all IBGP are fully mesh.
To overrule this rule, configure route reflector. Of course iBGP peer (R4) must know how to reach 10.12.1.2 and 10.13.1.2 via IGP or static routes.

------------

BGP Attributes

Well-known mandatory Attributes

AS-Path

Next-Hop; next-hop on shared media (if receiving router is on the same subnet, next-hop remains the same)

Origin - Tells how the route originated(IGP(I),EGP(E),Unknow(redistribution)).


Well-known optional Attributes

Local Preference - gives you control over preferred routes,higher is better.

Weight - cisco proprietary, gives you control of routes on the SAME router. It does not affect routing policy of other routers in the AS

Atomic aggregate - Informs router that a route has been summarized.

Multi-exit discriminator(MED) - used to suggest an entry point into your AS. Lower is better.

Aggregator - Designates the IP address of the router who performed summarization.

Community - Used for route tagging.

BGP Theory and basic config

Basics
Neighbors are manually configured.
Stage: IDLE, ACTIVE, Open Sent, Open Confirmed, Established

Hello sent every 60 seconds with a hold-down of 180 seconds.
Capable of MD5 authentication

Rule of Synchronization
Routes learnt via iBGP must be validated by the interior routing table before they can be advertised to remote peers - eBGP. Sync can be off/on at bgp router connected to remote ebgp peer.

Rule of Split-Horizon
Routes learnt via IBGP will never be sent to another IBGP peer
Route reflector to overcome this issue.


Attributes (Mandatory, Well-known Optional - transitive/non-transitive)
AS-Path, Next-Hop, Origin (IGP,EGP,Unknown?)
Local Preference(higher better), weight(local router, higher better) = select Exit point
Atomic Aggregate = this route is summarized
MED - used to suggest an entry point into your AS (lower is better)
Aggregator = designates IP addr of the router who performed summarization,
Community = used for route tagging

Basic Configuration
ip address of neighbors must be reachable(eg. via Internal routing protocol); exact subnet mask;
router bgp AS number
neighbor x.x.x.x remote-as 666
network 10.1.0.0 mask 255.255.0.0
neighbor 3.3.3.3 ebgp-multihop 2 -- if interfaces are not directly connected to each other
neighbor 3.3.3.3 update-source loopback 0 - to use if router's bgp address is loopback or not directly connected to neigbor
no auto summary

no synchronization -- turn of rule of synchronization
neighbor 2.2.2.2 next-hop-self -- when ebgp route is advertised back to ibgp

Troubleshooting
debug ip bgp updates