Interconnecting domains with Multiprotocol IBGPJuniper NetworksWienATkszarkowicz@juniper.netATT2212 Avenida MaraChula Vista91914CAUSisrael.means@att.comJuniper Networks18 Buckingham DrManalapan07726NJUSmnayman@juniper.net
General
IDR Working GroupInter-domainOption 10AOption 10BOption 10COption AOption BOption CMP-iBGPThis document relaxes the constraints specified in and allowing the building of Inter-domain L3VPN architecture with Multiprotocol internal BGP.IntroductionService provides must often partition (or divide) the large network into smaller IGP (Interrior Gateway Protocol) domains interconnected via BGP (Border Gateway Protocol) only. This might be required for various reasons; for example:
Separate geographic brown field networks: region 1, region 2, region 3 etc, for management or administrative purposes
Avoid advertising unnecessary routes from domain 1 to domain 2 to improve network scale of PE (Provider Edge) nodes and RR (Route Reflector) per region
Avoid advertising remote PE nodes loopback between regions, only DBR (Domain Boundary Router) nodes will advertise routes between regions using 'next-hop self' mechanism
The advantage of dividing the large network into smaller IGP domains can be numerous, with important examples like:
Per domain IGP (Interior Gateway Protocol) reduces blast radius during IGP errors or failures
Per domain RR reduces the blast radius and BGP message exchange when RR fails
At the same time, dividing the network can be impactful and result in unwanted behavior for both the operator and its customers. For example, some BGP attributes, such as LOCAL_PREF, are not sent to the EBGP (external BGP) peers but are sent to IBGP (internal BGP) peers. Also, depending on the actual requirements, operators can selectively choose, if they keep originator NEXT_HOP attribute or change the NEXT_HOP attribute to some local address. Further, Constrained Route Distribution () can be used to prevent DBR from sending VPN (Virtual Private Network) prefixes for VRFs (Virtual Routing and Forwarding instances) that are not locally attached to each region., in Section 10, describes three multi-domain L3VPN (Layer 3 Virtual Private Network) architectures - commonly referenced as Option 10A, Option 10B, and Option 10C - restricted to the use cases, where the domains are distinct BGP domains and use different AS (Autonomous System) numbers, therefore, these architectures use EBGP peerings between the domains. However, many operators might divide the network into multiple IGP domains keeping single BGP domain, with one AS number used across the IGP domains. This implies IBGP peers between IGP domains. In multi-domain architecture there might be a need to modify the NEXT_HOP path attribute at the domain boundary. While this is the default behavior for EBGP (, Section 5.1.3.), it is not recommended behavior for IBGP (, Section 10, recommends keeping NEXT_HOP path attribute unmodified when reflecting the NLRIs - Network Layer Reachability Information - between IBGP peers).This document relaxes these constraints specified in and , allowing Inter-domain L3VPN architectures stitching multiple IGP domains with MP-IBGP (Multiprotocol internal BGP).Requirements LanguageThe key words "MUST", "MUST NOT", "REQUIRED", "SHALL",
"SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT
RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be
interpreted as described in BCP 14 when, and only when, they appear in
all capitals, as shown here.Inter-domain L3VPN Option 10A with MP-IBGPInter-domain L3VPN architecture based on so called Option 10A (, Section 10, "a)" bullet point) relies on multiple logical interfaces (typically, sub-interfaces with unique VLAN - Virtual Local Area Network - per sub-interface) and multiple single-hop external BGP (SH-EBGP) peerings (single peering per sub-interface) between ASBRs (autonomous system boundary router), in an architecture as outlined in . Each SH-EBGP peering is responsible for exchanging unicast IPv4 (AFI/SAFI=1/1) or unicast IPv6 (AFI/SAFI=2/1) NLRIs for single L3VPN service. Essentially, in this architecture ASBRs consider each other as CE (Customer Edge) devices. RRs within each AS depicted in SHOULD be used. However, in small scale domains (for example, small access rings with few PEs), RR function could be placed on ASBRs, where multi-hop internal BGP (MH-IBGP) peerings are directly established between PEs and ASBRs.Inter-AS L3VPN Option 10A
SH-EBGP
MH-IBGP MH-IBGP SAFI=1 MH-IBGP MH-IBGP
SAFI=128 SAFI=128 ◀───────▶ SAFI=128 SAFI=128
◀───────▶ ◀───────▶ ◀───────▶ ◀───────▶ ◀───────▶
nhs nhs ◀───────▶ nhs nhs
nhs nhs
┌ ─ ─ ─ ─ ─ ─ ─ ─ ┌ ─ ─ ─ ─ ─ ─ ─ ─
┌──┐ │ ┌──┐ │
│ ╱│RR│╲ │ ╱│RR│╲
╱ └──┘ ╲ │ ╱ └──┘ ╲ │
┌┴─┐ ╱ ╲ ┌────┐ ┌──┴─┐ ╱ ╲ ┌──┐
│ │╱ ╲│ ├─────┤ │╱ ╲│ │
│PE│ AS 64501 │ASBR├─────┤ASBR│ AS 64502 │PE│
│ │ │ ├─────┤ │ │ │
└┬─┘ └────┘ └──┬─┘ └──┘
─ ─ ─ ─ ─ ─ ─ ─ ┘ ─ ─ ─ ─ ─ ─ ─ ─ ┘
This architecture does not require an end-to-end LSP (label switched path) leading from a packet's ingress PE in one AS to its egress PE in another AS, as the user packets exchanged between ASBRs are native IP (no MPLS - Multiprotocol Label Switching - encapsulation) packets. Hence, each ASBR has potentially multiple L3VPN service instances, and performs MPLS encapsulation/decapsulation, which is typical PE function. At the control plane level, each ASBR performs conversion between VPN-IPv4/VPN-IPv6 (SAFI=128) and unicast IPv4/IPv6 (SAFI=1) NLRIs. When these NLRIs are advertised by ASBR, NEXT_HOP attribute MUST be modified to self (nhs).In the original context described in , domains are BGP domains with different ASs, therefore, multiple BGP peerings between two BGP domains are EBGP. However, Option 10A concept can be applied not only to BGP domains with different AS numbers, but as well as to IGP domains with the same AS number, as depicted in .Inter-Domain L3VPN Option 10A using IBGP
MH-IBGP MH-IBGP SH-IBGP MH-IBGP MH-IBGP
SAFI=128 SAFI=128 SAFI=1 SAFI=128 SAFI=128
◀───────▶
◀───────▶ ◀───────▶ ◀───────▶ ◀───────▶ ◀───────▶
nhs nhs ◀───────▶ nhs nhs
nhs nhs
┌ ─ ─ ─ ─ ─ ─ ─ ─ ┌ ─ ─ ─ ─ ─ ─ ─ ─
┌──┐ │ ┌──┐ │
│ ╱│RR│╲ │ ╱│RR│╲
╱ └──┘ ╲ │ ╱ └──┘ ╲ │
┌┴─┐ ╱ ╲ ┌───┐ ┌─┴─┐ ╱ ╲ ┌──┐
│ │╱ ╲│ ├───────┤ │╱ ╲│ │
│PE│ AS 64500 │DBR├───────┤DBR│ AS 64500 │PE│
│ │ │ ├───────┤ │ │ │
└┬─┘ └───┘ └─┬─┘ └──┘
─ ─ ─ ─ ─ ─ ─ ─ ┘ ─ ─ ─ ─ ─ ─ ─ ─ ┘
The main differences, compared to the original Inter-domain Option 10A, are:
the BGP peering between two IGP domains are now IBGP (SAFI=1), and no longer EBGP (SAFI=1)
DBRs become PEs with all BGP peerings (global and inside VRFs) using the same AS number
Other aspects of the architecture are similar.Inter-domain L3VPN Option 10B with MP-IBGPInter-domain L3VPN architecture based on so called Option 10B (, Section 10, "b)" bullet point) relies on exchanging VPN-IPv4 (AFI/SAFI=1/128) or VPN-IPv6 (AFI/SAFI=2/128) NRLIs via direct SH-EBGP peering between ASBRs, in an architecture as outlined in . RRs within each AS depicted in SHOULD be used. However, in small scale domains (for example, small access rings with few PEs), RR function could be placed on ASBRs, where multi-hop internal BGP (MH-IBGP) peerings are directly established between PEs and ASBRs.Inter-AS L3VPN Option 10B
MH-IBGP MH-IBGP SH-EBGP MH-IBGP MH-IBGP
SAFI=128 SAFI=128 SAFI=128 SAFI=128 SAFI=128
◀───────▶ ◀───────▶ ◀───────▶ ◀───────▶ ◀───────▶
nhs nhs nhs nhs nhs nhs
┌ ─ ─ ─ ─ ─ ─ ─ ─ ┌ ─ ─ ─ ─ ─ ─ ─ ─
┌──┐ │ ┌──┐ │
│ ╱│RR│╲ │ ╱│RR│╲
╱ └──┘ ╲ │ ╱ └──┘ ╲ │
│ ╱ ╲ │ ╱ ╲
┌──┐╱ ╲┌─┴──┐ ┌────┐╱ ╲┌─┴┐
│PE│ │ASBR├─────┤ASBR│ │PE│
└──┘ AS 64501 └─┬──┘ └────┘ AS 64502 └─┬┘
└ ─ ─ ─ ─ ─ ─ ─ ─ └ ─ ─ ─ ─ ─ ─ ─ ─
This architecture requires an end-to-end LSP leading from a packet's ingress PE in one AS to its egress PE in another AS. Hence, at each ASBR, NEXT_HOP attribute MUST be modified to self (nhs), which results in new service label allocation, and programing of appropriate label forwarding entries in the data plane. On the ASBR-to-ASBR link between two ASs there is no additional 'labeled transport' (i.e., no LDP - Label Distribution Protocol, RSVP - Resource Reservation Protocol, SR - Segment Routing, ...) protocol - the packets are transmitted on the ASBR-to-ASBR link with single L3VPN service label.In the original context described in , domains are BGP domains with different ASs, therefore, the BGP peering between two BGP domains is EBGP. However, Option 10B concept can be applied not only to BGP domains with different AS numbers, but also to IGP domains with the same AS number, as depicted in .Inter-Domain L3VPN Option 10B using IBGP, with separate DBRs
MH-IBGP MH-IBGP SH-IBGP MH-IBGP MH-IBGP
SAFI=128 SAFI=128 SAFI=128 SAFI=128 SAFI=128
SAFI=132 SAFI=132 SAFI-132 SAFI=132 SAFI=132
◀───────▶ ◀───────▶ ◀───────▶ ◀───────▶ ◀───────▶
nhs nhs nhs nhs nhs nhs
┌ ─ ─ ─ ─ ─ ─ ─ ─ ┌ ─ ─ ─ ─ ─ ─ ─ ─
┌──┐ │ ┌──┐ │
│ ╱│RR│╲ │ ╱│RR│╲
╱ └──┘ ╲ │ ╱ └──┘ ╲ │
│ ╱ ╲ │ ╱ ╲
┌──┐╱ ╲┌─┴─┐ ┌───┐╱ ╲┌─┴┐
│PE│ │DBR├───────┤DBR│ │PE│
└──┘ AS 64500 └─┬─┘ └───┘ AS 64500 └─┬┘
└ ─ ─ ─ ─ ─ ─ ─ ─ └ ─ ─ ─ ─ ─ ─ ─ ─
Similar to the Inter-Domain Option 10A case, main differences, when comparing to the original Inter-domain Option 10B are:
the BGP peering between two IGP domains are now IBGP (SAFI=128), and no longer EBGP (SAFI=128)
DBRs become on-the-path route reflectors for SAFI=128
This implies that DBR with RR function MUST change the NEXT_HOP attribute to self, when reflecting these NLRIs. This is not in accordance with , Section 10 recommendation that RR SHOULD NOT modify the NEXT_HOP attribute, therefore, this document relaxes the recommendation from by defining the use case, where RR MUST modify the NEXT_HOP attribute, when reflecting NRLIs over IBGP peerings.It is strongly advisable to control the exchange of VPN-IPv4/VPN-IPv6 (SAFI=128) NLRIs between domains via Constrained Route Distribution (). Therefore, DBR-to-DBR SH-IBGP peering, in addition to SAFI=128, SHOULD include Route Target Constraint - RTC (SAFI=132) - as well, and DBRs SHOULD be provisioned to exchange between each other only desired RTCs. Please note, RTC MAY be used inside of each IGP domain, too, to control route distribution within IGP domains.Important aspect of the inter-domain conenctivity, when the domains are interconnected via multiple interconnection points, as depicted in , is loop prevention. In classic Inter-AS Option 10B, with different AS numbers used in each BGP domain, and EBGP peerings between BGP domains, loop prevention is ensured by rejecting updates containig local AS number in the AS_PATH attribute. In the use case with multiple IGP domains and single BGP domain, each DBR is on-the-path RR, thus is associated with a CLUSTER_ID. One option for CLUSTER_ID alloaction is, that each DBR is configured with a unique CLUSTER_ID. Another option for CLUSTER_ID alloaction is that each DBR pair in the IGP domain uses unique CLUSTER_ID, as depicted in . Using the first CLUSTER_ID allocation scheme, there is a risk of a BGP routing loop occurring, since all of the DBRs are reflecting prefixes as RRs in the same AS with next-hop self. To prevent the DBRs of the same IGP domain from accepting updates from each other, they SHOULD use the same CLUSTER_ID. In this case, a DBR will discard a prefix update that has the same CLUSTER_ID as itself in order to prevent routing information loops in BGP. For example, DBR1 and DBR2 configured with one CLUSTER_ID, while DBR3 and DBR4 have another single CLUSTER_ID. This mechanism, loop prevention based on CLUSTER_LIST filtering, is described in , Section 10 - Avoiding Routing Information Loops.Loop prevention in Inter-Domain L3VPN Option 10B using IBGP
MH-IBGP MH-IBGP SH-IBGP MH-IBGP MH-IBGP
SAFI=128 SAFI=128 SAFI=128 SAFI=128 SAFI=128
SAFI=132 SAFI=132 SAFI-132 SAFI=132 SAFI=132
◀───────▶ ◀───────▶ ◀───────▶ ◀───────▶ ◀───────▶
nhs nhs nhs nhs nhs nhs
┌ ─ ─ ─ ─ ─ ─ ─ ─ ┌ ─ ─ ─ ─ ─ ─ ─ ─
┌──┐ ┌─┴──┐ ┌────┐ ┌─┴┐
│PE│ │DBR1│───────┤DBR3│ │PE│
└──┘ └─┬──┘ └────┘ └─┬┘
│ ╲ ╱ C C │ ╲ ╱
╲ ╱ │ l l ╲ ╱ │
│ ╲ ╱ u u │ ╲ ╱
╲┌──┐╱ │ s s ╲┌──┐╱ │
│ │RR│ t t │ │RR│
╱└──┘╲ │ e e ╱└──┘╲ │
│ ╱ ╲ r r │ ╱ ╲
╱ ╲ │ ╱ ╲ │
│ ╱ ╲ A B │ ╱ ╲
┌──┐ ┌─┴──┐ ┌────┐ ┌─┴┐
│PE│ AS 64500 │DBR2├───────┤DBR4│ AS 64500 │PE│
└──┘ └─┬──┘ └────┘ └─┬┘
└ ─ ─ ─ ─ ─ ─ ─ ─ └ ─ ─ ─ ─ ─ ─ ─ ─
When using IBGP, instead of EBGP, small variation of the architecture can be achieved, by collapsing two separate DBRs to single, collapsed DBR, as depicted in .Inter-Domain L3VPN Option 10B using IBGP, with collapsed DBR
MH-IBGP MH-IBGP MH-IBGP MH-IBGP
SAFI=128 SAFI=128 SAFI=128 SAFI=128
SAFI=132 SAFI=132 SAFI=132 SAFI=132
◀───────▶ ◀───────▶ ◀───────▶ ◀───────▶
nhs nhs nhs nhs
┌ ─ ─ ─ ─ ─ ─ ─ ┐ ┌ ─ ─ ─ ─ ─ ─ ─ ┐
┌──┐ ┌──┐
│ ╱│RR│╲ │ │ ╱│RR│╲ │
╱ └──┘ ╲ ╱ └──┘ ╲
│ ╱ ╲ │ │ ╱ ╲ │
┌──┐╱ ╲┌─────┐╱ ╲┌──┐
│PE│ │ DBR │ │PE│
└──┘ AS 64500 └─────┘ AS 64500 └──┘
└ ─ ─ ─ ─ ─ ─ ─ ┘ └ ─ ─ ─ ─ ─ ─ ─ ┘
Similarly to the previous example, DBR MUST change the NEXT_HOP attribute to self, when reflecting VPN-IPv4/VPN-IPv6 (SAFI=128) NLRIs, and DBR SHOULD use RTC (SAFI=132) to control the exchange of VPN-IPv4/VPN-IPv6 (SAFI=128) NLRIs between domains. RTC MAY be used inside of each domain.Inter-domain L3VPN Option 10C with MP-IBGPInter-domain L3VPN architecture based on so called Option 10C (, Section 10, "c)" bullet point) relies on exchanging VPN-IPv4 (AFI/SAFI=1/128) or VPN-IPv6 (AFI/SAFI=2/128) NLRIs via MH-EBGP peering between BGP domains, without changing the NEXT_HOP attribute, and exchanging labeled unicast IPv4 or labeled unicast IPv6 (SAFI=4) host routes (PE loopbacks) via direct SH-EBGP peering between ASBRs, changing the NEXT_HOP attribute at the BGP domain boundaries, in an architecture as outlined in . As in previous architectures, RRs within each AS depicted in SHOULD be used. One of the main objectives of Option 10C architecture is to offload ASBRs from the task of maintaining/distributing VPN-IPv4/VPN-IPv6 (SAFI=128) NLRIs, without RR these NLRIs would need to be distributed via direct MH-EBGP peerings between PEs from different BGP domains. Such approach makes the design very impractical and not scalable, therefore, in Option 10C RRs SHOULD be deployed, and MH-EBGP peerings to distribute VPN-IPv4/VPN-IPv6 (SAFI=128) NLRIs between BGP domains SHOULD be established between RRs.Inter-AS L3VPN Option 10C
MH-IBGP MH-EBGP MH-IBGP
SAFI=128 SAFI=128 SAFI=128
◀───────▶ ◀───────────────────────────▶ ◀───────▶
nhs nhs
MH-IBGP SH-EBGP MH-IBGP
SAFI=4 SAFI=4 SAFI=4
◀─────────────────▶ ◀───────▶ ◀─────────────────▶
nhs nhs nhs nhs nhs nhs
┌ ─ ─ ─ ─ ─ ─ ─ ─ ┌ ─ ─ ─ ─ ─ ─ ─ ─
┌──┐ │ ┌──┐ │
│ ╱│RR│╲ │ ╱│RR│╲
╱ └──┘ ╲ │ ╱ └──┘ ╲ │
│ ╱ ╲ │ ╱ ╲
┌──┐╱ ╲┌─┴──┐ ┌────┐╱ ╲┌─┴┐
│PE│ │ASBR├─────┤ASBR│ │PE│
└──┘ AS 64501 └─┬──┘ └────┘ AS 64502 └─┬┘
└ ─ ─ ─ ─ ─ ─ ─ ─ └ ─ ─ ─ ─ ─ ─ ─ ─
This architecture requires an end-to-end LSP leading from a packet's ingress PE in one AS to its egress PE in another AS. Hence, at each ASBR, NEXT_HOP attribute for labeled unicast IPv4 or labeled unicast IPv6 (SAFI=4) NLRI MUST be modified to self (nhs), which results in new transport label allocation, and programming of appropriate label forwarding entries in the data plane. In the packets traversing ASBR-to-ASBR link between two ASs, similar to the links within each AS, there is additional transport label at the top of the label stack in addition to the L3VPN service label. This transport label is exchanged via BGP peering with SAFI=4.In the original context described in , domains are BGP domains with different AS numbers, therefore, the BGP peerings (both for SAFI=4 and SAFI=128) between two BGP domains are EBGP. However, Option 10C concept can be applied not only to BGP domains with different AS numbers, but as well to IGP domains with the same AS number, as depicted in .Inter-Domain L3VPN Option 10C using IBGP, with separate DBRs
MH-IBGP MH-IBGP MH-IBGP
SAFI=128 SAFI=128 SAFI=128
SAFI=132 SAFI=132 SAFI=132
◀───────▶ ◀───────────────────────────▶ ◀───────▶
nhs nhs
MH-IBGP SH-IBGP MH-IBGP
SAFI=4 SAFI=4 SAFI=4
◀─────────────────▶ ◀───────▶ ◀─────────────────▶
nhs nhs nhs nhs nhs nhs
┌ ─ ─ ─ ─ ─ ─ ─ ─ ┌ ─ ─ ─ ─ ─ ─ ─ ─
┌──┐ │ ┌──┐ │
│ ╱│RR│╲ │ ╱│RR│╲
╱ └──┘ ╲ │ ╱ └──┘ ╲ │
│ ╱ ╲ │ ╱ ╲
┌──┐╱ ╲┌─┴─┐ ┌───┐╱ ╲┌─┴┐
│PE│ │DBR├───────┤DBR│ │PE│
└──┘ AS 64500 └─┬─┘ └───┘ AS 64500 └─┬┘
└ ─ ─ ─ ─ ─ ─ ─ ─ └ ─ ─ ─ ─ ─ ─ ─ ─
Again, the differences compared to the original Inter-domain Option 10C are:
the peerings between two IGP domains are now IBGP, and no longer EBGP, for both single-hop BGP peering used to exchange labeled unicast IPv4 or labeled unicast IPv6 (SAFI=4) host routes (PE loopbacks), as well as multi-hop BGP peering used to exchange VPN-IPv4 (AFI/SAFI=1/128) or VPN-IPv6 (AFI/SAFI=2/128) NLRIs
DBRs become on-the-path route reflectors for SAFI=4
Remaining aspects of the architecture are similar. This implies that IGP domain boundary router (DBR) becomes inline (on-the-path) RR for labeled unicast IPv4 or labeled unicast IPv6 (SAFI=4) NLRIs, and MUST change the NEXT_HOP attribute to self, when reflecting these NLRIs. Again, this is not in accordance with , Section 10 recommendation that RR SHOULD NOT modify the NEXT_HOP attribute, therefore, this document relaxes the recommendation from by defining the use case, where RR MUST modify the NEXT_HOP attribute, when reflecting NRLIs over IBGP peeringsAs in Option 10B scenario, it is strongly advisable to control the exchange of VPN-IPv4/VPN-IPv6 (SAFI=128) NLRIs between domains via Constrained Route Distribution (). Therefore, MH-IBGP peering between RRs in different IGP domains, in addition to SAFI=128, SHOULD include RTC (SAFI=132), and RRs SHOULD be provisioned to exchange between each other only desired RTCs. Please note, RTC MAY be used inside of each domain, too, to control route distribution within IGP domains.When using IBGP, instead of EBGP, a small variation of the architecture can be achieved, by collapsing two separate DBRs to single, collapsed DBR, as depicted in .Inter-Domain L3VPN Option 10C using IBGP, with collapsed DBR
MH-IBGP MH-IBGP MH-IBGP
SAFI=128 SAFI=128 SAFI=128
SAFI=132 SAFI=132 SAFI=132
◀───────▶ ◀─────────────────▶ ◀───────▶
nhs nhs
MH-IBGP MH-EBGP
SAFI=4 SAFI=4
◀─────────────────▶ ◀─────────────────▶
nhs nhs nhs nhs
┌ ─ ─ ─ ─ ─ ─ ─ ┐ ┌ ─ ─ ─ ─ ─ ─ ─ ┐
┌──┐ ┌──┐
│ ╱│RR│╲ │ │ ╱│RR│╲ │
╱ └──┘ ╲ ╱ └──┘ ╲
│ ╱ ╲ │ │ ╱ ╲ │
┌──┐╱ ╲┌─────┐╱ ╲┌──┐
│PE│ │ DBR │ │PE│
└──┘ AS 64500 └─────┘ AS 64500 └──┘
└ ─ ─ ─ ─ ─ ─ ─ ┘ └ ─ ─ ─ ─ ─ ─ ─ ┘
Similarly to the previous example, DBR MUST change the NEXT_HOP attribute to self, when reflecting labeled unicast IPv4 or labeled unicast IPv6 (SAFI=4) NLRIs, and RR SHOULD use RTC (SAFI=132) to control the exchange of VPN-IPv4/VPN-IPv6 (SAFI=128) NLRIs between IGP domains. RTC MAY be used inside of each domain.IANA ConsiderationsThis memo includes no request to IANA.Security ConsiderationsTo be added later.ReferencesNormative ReferencesInformative ReferencesAcronyms and AbbreviationsAFI: Address Family IdentifierAS: Autonomous SystemASBR: Autonomous System Boundary RouterBGP: Border Gateway ProtocolCE: Customer EdgeDBR: Domain Boundary RouterEBGP: External Border Gateway ProtocolIBGP: Internal Border Gateway ProtocolIGP: Interior Gateway ProtocolIP: Internet ProtocolIPv4: Internet Protocol version 4IPv6: Internet Protocol version 6L3VPN: Layer 3 Virtual Private NetworkLDP: Label Distribution ProtocolLSP: Label Switched PathMH-IBGP: Multi-hop Internal Border Gateway ProtocolMP-IBGP: Multiprotocol Internal Border Gateway ProtocolMPLS: Multiprotocol Label Switchingnhs: next-hop selfNLRI: Network Layer Reachability InformationPE: Provider EdgeRR: Router ReflectorRSVP: Resource Reservation ProtocolRTC: Route Target ConstraintSAFI: Subsequent Address Family IdentifierSH-EBGP: Single-hop External Border Gateway ProtocolSR: Segment RoutingVLAN: Virtual Local Area NetworkVPN: Virtual Private NetworkVRF: Virtual Routing and ForwardingAcknowledgementsTo be added laterContributorsTo be added later