What is MPLS?

MPLS (Multiprotocol Label Switching) is a packet-forwarding technique used inside operator networks where the forwarding decision is made on a short label rather than a full IP destination lookup. An edge router attaches a fixed-length MPLS header to the packet, and the routers in the core switch traffic purely by that label, swapping it at every hop. The core therefore does not need to carry the full internet routing table, and the operator gets one consistent foundation for Layer 3 VPNs, traffic engineering and fast failover.

How label switching works

The MPLS architecture is defined in RFC 3031. The label is a 20-bit field inside a 32-bit MPLS header that also carries three Traffic Class bits (formerly EXP), a Bottom of Stack bit and an 8-bit TTL field. Edge routers (LERs) classify packets into FECs and impose the first label, while core routers (LSRs) perform swap, push and pop operations as traffic travels along a label switched path (LSP). Labels are distributed by signalling such as LDP (RFC 5036) or RSVP-TE. The last hop often uses penultimate hop popping, which removes the transport label just before the egress router.

• Push: the edge router imposes a label onto an incoming IP packet.
• Swap: a core router replaces the incoming label with an outgoing one and forwards the packet.
• Pop: the label is removed, usually via penultimate hop popping before the egress router.
• Label stack: MPLS lets you stack multiple labels, so VPNs and traffic engineering run at the same time.
• FEC: a Forwarding Equivalence Class groups packets treated identically along the whole LSP.

L3VPN and customer separation

The most common use of MPLS is the Layer 3 VPN defined in RFC 4364 (formerly known as RFC 2547bis). Each customer gets its own VRF table on the PE router, so address space can overlap between customers without collisions. Customer routes are carried as the VPNv4 or VPNv6 family in MP-BGP, an 8-byte Route Distinguisher keeps them unique, and Route Targets control which routes are imported and exported between VRFs. Transport uses a stack of two labels: the outer one carries the packet across the core to the correct PE, while the inner one identifies the destination VRF or customer interface.

Traffic engineering and resilience

MPLS Traffic Engineering lets operators steer LSPs along paths other than the shortest IGP metric would choose, routing traffic over links with the right capacity. TE tunnels signalled with RSVP-TE can reserve bandwidth and avoid congested segments, and in newer networks Segment Routing over MPLS (SR-MPLS) does similar work while removing per-LSP signalling state from the core. For resilience, fast-restoration mechanisms such as MPLS Fast Reroute switch traffic to a backup path within tens of milliseconds of a link or node failure.

MPLS and the AS202520 SkyPass network

AS202520 SkyPass runs a core linking its PoPs in Warsaw and Wrocław with predictable, low-latency transport of customer traffic in mind. Most customers care primarily about BGP-based services: IP transit, peering and remote access to Polish internet exchanges such as THINX, TPIX, WRIX and 1-IX. In modern carrier networks, traffic isolation, path engineering and fast failover between sites come from MPLS-style and Segment Routing techniques. If you are planning multihoming or dedicated transport between Warsaw and Wrocław, our team will help you choose the right connectivity model.