Transcript
Overview
}Introduction:
The past
MPLS The solution }One MPLS network for all services
L2 Ethernet Services
VPN IP
3Play
ATMoMPLS }Lessons Learned
FRR
OAM
LACP
Loop protection on VPLS Access
Service provisioning
OSS
OE/OM }Conclusions
PT’s Ethernet Network Limitations
. Limited number of logical connections .Maximum VLAN number = 4093 .VLAN ID global significance
. MAC Explosion .MAC learning required in all services .Seconds Convergence time (spanning tree protocols)
.Operational problems .No OAM tools .Loops not allowed
.Poor QOS .Based on 802.1p
Requirements for new
MPLS based Network
9 Increased number of logical connections 9VLAN ID local significance 9 DSLAM aggregation .VLAN encapsulation flexibility
required
9 Restoration time ms
9FRR 9 No MAC learning (point-to-point connection) 9More efficient load balancing on Network links 9Efficient QoS 9Traffic Engineering 9OAM Tools
9MAC Ping, MAC Trace, …
One Network for all Services
}L2 Ethernet Services DSL Services Business Services
}VPN IP }3Player }ATMoMPLS
}Not directly Frame Relay TDM
Business L2 Ethernet Services
EPL/EVPL E-LAN
Site 1 Site 3
MPLS Network
Site 2 Site 4
Business L2 Ethernet Services
} COS } SLAs } Flexible Bandwidth } Costumer Portal } Reporting } Managed CPEs } Access Network Resilience } MEF 9 . Carrier Ethernet Certified
Access Protection
MPLS
Network
2 POPs 2 Diverse Routes 2 Interfaces
Triple Play Aggregation Network
BSA BSR
32 11
ST
B
ST B RG DSLAM BTV
PC MPLS
PC Network
VOD
32 11
ST MPLS
B ST RG ETHERNET VoIP
B
PC
PC DSLAM
33 11
ST B BSR
ST RG 32 9
B
PC 33 9
PC
BRAS
16 Internet
FRR issues
}Experienced Inter-Vendor interoperability issues FRR Misunderstood messages “Zombie” Detours CPU utilization (mainly because CSPF calculations) Service affecting (in some cases)
}When we started (2005) Facility Mode wasn’t operable between two vendor’s routers Started with One-to-One Mode } Today already using Facility Mode But still problems with “Global Revertive Mode” behaviour of the PLR
OAM Fault Management issues
} Sub 50 ms restoration (FRR) but Tens of seconds to detect RSVP session or IGP adjacency failed }Inexistent Link fault detection mechanisms
Need for network links over active transport networks
EoSDH
EoC/DWDM
}Although available BFD is not “the” solution Not supported on LACP Not supported on LR
LACP Hashing issues
} LACP Hashing algorithm only taking in account tunnel and service labels Non optimal link utilization Impossible having one service with more than 1G traffic } Other parameters are required for hashing, L2/L3/L4 address/ports not yet supported as a basic feature
Evolution of PT’s Network
} MPLS Network
High Availability
ISSU
LSP point-to-multipoint
Additional fault management functions
IEEE 802.3ah -EFM
IEEE 802.1ag -CFM
MEF OAM
ITU Y.1731
Multi-chassis LACP
} Transport Network Multicast support OTN T-MPLS GPON PBB-TE
} Access Network EDD
Standard compliance for multi-vendor interoperability Advanced OAM capabilities Testing and loopback capabilities
RFC2544 Link and Connectivity management Cable integrity testing
TDR support Optical Performance Monitoring End-to-End Performance Monitoring
Conclusions
}We are currently deploying a successful MPLS Networks to offer Business and Residential Triple-Play services
}Our OSS/BSS is a key factor for the success and evolution of the network and the services provided
}There is still work to be done to improve multi-vendor full MPLS
interoperability
}More OAM tools are required