In an IES service, the router acts as a host. The most straightforward and standard design is to create separate interfaces for the WAN link (point-to-point /30) and the customer LAN (broadcast /24). The DHCP server can be enabled directly on the LAN interface to serve addresses from its configured subnet. This keeps the services logically separate, follows the typical hub-and-spoke model for internet access, and is the recommended practice in the Nokia service architecture.

While RDs and RTs are critical for MP-BGP route distribution, they do not affect the underlying OSPF neighbor adjacency formation between the PE and CE. OSPF adjacency is established at the interface level within the VPRN context. A common, often overlooked issue is the service MTU on the SAP (Service Access Point). If the configured service MTU is smaller than the interface MTU, OSPF hello packets (which can be large with options) may be silently dropped, preventing the adjacency from forming. This is a practical, real-world troubleshooting step.

In H-VPLS, spoke-SDPs are typically used in a hub-and-spoke topology towards aggregation nodes. The split-horizon group is a fundamental loop prevention mechanism. When applied to a group of spoke-SDPs, it ensures that if a broadcast, unknown unicast, or multicast (BUM) frame is received from one spoke-SDP in the group, it will not be forwarded out to any other spoke-SDP within the same group. This prevents loops in the layer 2 domain without requiring a full mesh of pseudowires or running a spanning tree protocol over the SDPs.

In SR-MPLS, the ingress node (or a controller) builds the label stack. The destination prefix is associated with a Prefix-SID (Node-SID), which becomes the bottom label, ensuring the packet reaches the destination node. If an explicit path (e.g., for traffic engineering) is required, the ingress node pushes additional labels on top, such as Adjacency-SIDs, to specify the exact links or nodes the packet must traverse. The top label in the stack is the one that corresponds to the first segment/instruction.

When a BGP session is in the ‘Active’ state, the local router is trying to initiate a TCP connection. If the configured ‘local-address’ (the source IP for the BGP TCP session) is not an IP address that is active and operational on a local interface, the TCP SYN packet cannot be sent. The router will continuously retry, keeping the session in the Active state. This is a very specific and common configuration mistake, especially when using loopback addresses for peering that haven’t been correctly assigned to the system interface.

In mLDP, the control plane uses downstream-on-demand label distribution for P2MP LSPs. The process is receiver-initiated. A leaf node (the receiver) sends an mLDP Label Request message for a specific P2MP FEC (identifying the tree) upstream towards the root. This request travels hop-by-hop along the reverse path. Each upstream node, upon receiving the request from a downstream node, creates state and forwards its own request further upstream. Eventually, the root responds with a Label Mapping, and labels are distributed down the tree, establishing the P2MP LSP.

By default, IS-IS uses a narrow metric (6-bit) with a maximum link cost of 63, which often makes all high-speed links appear equal (cost of 10). Widening the reference-bandwidth (or using wide metrics) changes the calculation so that the default cost is (reference-bandwidth / link-bandwidth). This provides a much more granular metric range (24-bit), allowing higher bandwidth links to receive a lower, more attractive cost. This directly influences the SPF calculation, making the protocol prefer these links as the true shortest path, which is the most natural way to achieve the goal.

Within a VPRN, customer routes (learned from the CE) have next-hops that are other PE routers (for inter-site traffic) or the internet gateway. These next-hops are in the service provider’s core address space. The VPRN must have a way to resolve these BGP next-hops to an MPLS tunnel (LSP). The ‘auto-bind’ command (e.g., `auto-bind lsp`) in the VPRN configuration tells the router to use LSPs from the global routing table to reach these next-hops. If this is missing or misconfigured, the routes will be in the RIB but unresolvable in the forwarding plane, causing the described symptom.

This describes a requirement for a multi-chassis active/standby scenario where the entire service should be presented from one chassis or the other, not load-balanced. Placing the two SAPs (on different chassis) in the same Multi-chassis Synchronization (MCS) group allows them to synchronize state. In such a group, only the SAP on the active chassis is in a forwarding state. If the active chassis or its link fails, the standby SAP is activated. Crucially, if the active SAP’s physical port fails, the MCS protocol will bring down the *service* on the active chassis, triggering a switchover to the standby, rather than trying to use the other local port. This ensures a clean, all-or-nothing failover.

In an IES service, the router acts as a host. The most straightforward and standard design is to create separate interfaces for the WAN link (point-to-point /30) and the customer LAN (broadcast /24). The DHCP server can be enabled directly on the LAN interface to serve addresses from its configured subnet. This keeps the services logically separate, follows the typical hub-and-spoke model for internet access, and is the recommended practice in the Nokia service architecture.0