Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
MPLS architectures
Seamless, scalable, resilient MPLS
for mission-critical networks
Contents
MPLS: A proven technology
11
11
Conclusion 14
Acronyms 15
Remote
location #1
B
Remote
location #2
Aggregation
location #1
Core
Domain
Central
location
NOC #1
Aggregation
Domains
Aggregation
location #2
Central
location
NOC #2
Remote
location #3
A single end-to-end MPLS domain: Seamless MPLS extends the core domain
and integrates aggregation domains into a single MPLS domain. This single
domain enables more efficient management and troubleshooting of the
transport and services layers.
A network without boundaries (seamless): Seamless MPLS allows MPLSbased services to be established between any two endpoints without
per-service configuration in intermediate nodes at aggregation locations.
Rapid fault detection and recovery: Seamless MPLS supports end-to-end
fault detection, fast protection and end-to-end operations, administration
and maintenance (OAM) functions.
Decoupling of transport layer from services: Seamless MPLS allows services
to be provisioned wherever they are needed, regardless of the architecture
of the underlying transport layer. This is achieved by implementing a
three-label hierarchy consisting of a transport layer and a services layer
(see Figure 2).
Figure 2. Decoupling transport layer from services
MPLS
Aggregation
Domain
MPLS
Core
Domain
PE-11
iBGP
PE-12
IP/MPLS
Aggregation
Domain
P1
(RR)
PE-21
Central
location
NOC #1
iBGP
IP/MPLS
Core Domain
iBGP
ABR-12
(RR)
ABR-22
PE-22
Central
location
NOC #2
ABR-11
(RR)
PE-11
VLL
VLL
iBGP
IP/MPLS
Aggregation
Domain
PE-12
PE-21
Central
location
NOC #1
iBGP
IP/MPLS
Core Domain
iBGP
IP-VPN
IP-VPN
ABR-12
(RR)
ABR-22
PE-22
Central
location
NOC #2
Decoupling the transport and services layers within the Seamless MPLS
framework allows services to be provisioned wherever they are needed,
independent of the underlying transport layer. As more services are required
between PE-11 and PE-21 or PE12 and PE-22, only T-LDP and MP-BGP4 are
used to signal new service label to create new service binding directly between
them. There is no involvement of the transport tunnel and intermediate
nodes, including the RR.
PW redundancy
Transport
level
MPLS FRR
MPLS FRR
P1
ABR-11
(RR)
PE-11
VLL
VLL
iBGP
IP/MPLS
Aggregation
Domain
PE-12
PE-21
Central
location
NOC #1
iBGP
IP/MPLS
Core Domain
iBGP
IP-VPN
IP-VPN
ABR-12
(RR)
ABR-22
PE-22
Central
location
NOC #2
PE1
iBGP
BGP
prexes
PE1
BGP NH1
IGP NH1
oif
BGP NH2
IGP NH2
oif
SET 1
iBGP
10
MPLS FRR can be accomplished using two methods, both of which can be used
to protect links and nodes during network failure:
The one-to-one backup method, which creates detour LSPs for each
protected LSP at each potential point of local repair
The facility backup method, which creates a bypass tunnel to protect a
potential failure point; by taking advantage of MPLS label stacking, this
bypass tunnel can protect a set of LSPs that have similar backup constraints
11
area LSPs)4. RFC 5283 defines a new LDP label mapping procedure to support
setting up contiguous inter-area LSPs while maintaining IP prefix aggregation
on the ABR nodes.
This procedure is similar to the one defined in the LDP specification (RFC
5036) but performs an IP longest-match lookup when searching the Forward
Equivalence Class (FEC) element in the Routing Information Base (RIB).
Pros
LSP transport tunnels can be created endto-end.
Services can be deployed without provisioning intermediate points.
With prefix aggregation, leaking all the /32 FECs into the IGP area/level is
not required, thereby reducing routing table size.
Cons
If prefix aggregation (summarized entries) is not supported, the router table
sizes can become quite large, burdening router resources. In addition, large
tables take longer to converge and make troubleshooting complex.
Even if prefix aggregation is supported and IP Routing Information Base/
IP Forwarding Information Base (IP RIB/IP FIB) tables are reduced, Label
Forwarding Information Bases (LFIBs) are still flooded with all the /32 FECs
of the whole network, reducing the overall scalability and increasing the
complexity.
Reducing FEC distribution requires complex policies.
12
Pseudowire switching
Pseudowire switching (also called multi-segment PW) allows VLL services
to be scaled over a multi-area network by making a full mesh of TargetedLDP sessions between PE nodes unnecessary. The end-to-end segment is
split into multiple segments that are switched at switching points. PE nodes
that terminate the end-to-end service are referred to as T-PEs and the
intermediate PEs at the junctions of each segment are referred to as S-PEs.
The T-PE node acts as a master and the S-PE nodes act as slaves for PW
signaling. The S-PE waits for an LDP-mapping message from the T-PE. The
PW is signaled using T-LDP. Pseudowire switching limits the propagation of
/32 FECs; however PW switching requires provisioning at multiple points
(T-PE and S-PEs).
Pros
A full mesh of T-LDP sessions between PE nodes is not required.
Propagation of /32 FECs and router table size are limited.
Cons
You may need to provision intermediate points unless dynamic multisegment PW (draft-ietf-pwe3-dynamic-ms-pw) is deployed in every
T-PE and S-PE, which adds another layer of complexity.
End-to-end troubleshooting is more complex.
No L3 VPN is possible because only VLL services are supported.
Inter-AS options
RFC 4364 (BGP/MPLS IP VPNs)5 describes three options for supporting interAS IP-VPNs.
Option A uses back-to-back connections between the AS boundary router
(ASBR) nodes. This option does not support end-to-end MPLS and is only
suitable when the number of IP-VPNs is very small because the option
requires per-VPN configuration on ASBRs (that is, a sub-interface and external
BGP (eBGP) session is required for each IP-VPN).
Option B eliminates the need for per-VPN configuration on the ASBRs. The
ASBRs receive IP-VPN information from PEs in the local AS and forward this
information to their eBGP peer ASBRs. Each peer ASBR, in turn, forwards the
IP-VPN information to its local BGP peers in the remote AS. This option is
suitable for inter-AS IP-VPNs when working with different service providers
because all routes advertised between ASs can be controlled by routing
policies at the ASBR.
Option C is essentially the seamless MPLS approach to implementing interAS VPRNs. With Option C, VPN prefixes are neither held nor re-advertised by
5
13
the ASBR. PEs in different ASs can establish multi-hop, multi-protocol eBGP
sessions with each other to exchange customer VPN prefixes. Together with
the inter- and intra-domain transport tunnels previously described, a threelevel label stack is formed, which is essentially the Seamless MPLS architecture
model.
The bottom-level label is assigned by the egress PE (advertised in multi-hop,
multi-protocol eBGP without next-hop override) and is commonly referred
to as the VPN label or service label. The middle label is assigned by the local
ASBR-PE and corresponds to the /32 route of the egress PE (in a different AS)
using BGP-LBL (RFC 3107, Carrying Label Information in BGP-4). The top-level
label is assigned by the local ASBR-PE(s)/32 loop-back address, which would
be assigned by the IGP next-hop of the ingress PE.
Option C allows for a higher scale of VPRNs across AS boundaries and also
expands the trust model between ASs. As a result, this model is typically
used within a single company that may have multiple ASs.
Conclusion
MPLS is the preferred technology to implement scalable networks for missioncritical networks as well as service provider networks carrying business,
residential and mobile services. Although several options and technologies can
be used to implement end-to-end MPLS networks, Seamless MPLS provides
maximum scalability, flexibility and ease of provisioning and maintenance.
Some networks may not be ready to implement Seamless MPLS architectures
immediately. Alternative approaches described at the end of this paper can
be considered as interim solutions to deploy end-to-end MPLS networks and
services. For a fully scalable and flexible solution, however, the Seamless MPLS
framework must be considered in the future.
Nokia, a leader in MPLS development and IP service routing, offers
a complete, comprehensive and industry-validated toolkit that enables
network operators to migrate to or implement new end-to-end MPLS
network architectures.
14
Acronyms
ABR
LSP
label-switched path
ATM
MP-BGP
Multiprotocol BGP
AS
autonomous system
MPLS
ASBR
NOC
BFD
OAM
BGP
OSPF
CFM
PDH
eBGP
external BGP
PE
Provider Edge
FEC
PIC
Prefix-Independent Convergence
FIB
PW pseudowire
FRR
Fast Reroute
RIB
iBGP
IETF
IGP
IP
Internet Protocol
IP FIB
IP RIB
RR
Route Reflector
SAA
SCADA
SDH
SONET
T-LDP
LAN
VLL
LDP
LFIB
Nokia is a registered trademark of Nokia Corporation. Other product and company names
mentioned herein may be trademarks or trade names of their respective owners.
Nokia Oyj
Karaportti 3
FI-02610 Espoo
Finland
Tel. +358 (0) 10 44 88 000
Product code: PR1605020293EN (July)
Nokia 2016
nokia.com