Layer 2 VLAN Traffic Engineering MPLS

[walliaralot]

Trying to find information on how to configure Traffic Engineering, or something like rate limit, in a layer 2 VLAN EoMPLS solution.

3800--layer2-VLAN---6509-------MPLS--------6509--layer2-VLAN--3800

thanks, Kevin

[Roman Rodichev]

Traffic Engineering banwidth reservation is only logical and doesn't do any kind of rate limiting

Policing for MPLS traffic with Layer 2 PDU inside is also not possible (it would be possible if it was L3VPN)

Consider switching from EoMPLS to EoIP (L2TPv3), you can then configure a regular QoS policy-map that polices the L2TPv3 traffic between your 6500 loopbacks

[walliaralot]

okay, makes sense.

wouldn't happen to know of a good white-page for EoIP?

thanks!

[Roman Rodichev · Accepted solution]

here is your whitepaper

6509-1

int loopback0

ip address 10.0.0.1

!

pseudowire-class EOIP

encapsulation l2tpv3

interworking ip    <<<< OPTIONAL if you have different L2 types on two sides, don't need it if it's VLAN mode on both sides

ip local interface Loopback0

!

interface Vlan1000

xconnect 10.0.0.2 999 pw-class EOIP

!

!

ip access-list ex EOIP

permit 115 host 10.0.0.1 host 10.0.0.2          <<<< IP Protocol 115 is used by L2TPv3 for encapsulation

!

class-map EOIP

match access-group name EOIP

!

policy-map QOS

class EOIP

  police 1000000               <<<<< Police to 1Mbps

!

interface Gig1/1 (or whatever interface facing core)

ip address x.x.x.x x.x.x.x

service-policy output QOS

6509-2

int loopback0

ip address 10.0.0.2

!

pseudowire-class EOIP

encapsulation l2tpv3

interworking ip

ip local interface Loopback0

!

interface Vlan1000

xconnect 10.0.0.1 999 pw-class EOIP

!

!

ip access-list ex EOIP

permit 115 host 10.0.0.2 host 10.0.0.1

!

class-map EOIP

match access-group name EOIP

!

policy-map QOS

class EOIP

  police 1000000

!

interface Gig1/1

ip address x.x.x.x x.x.x.x

service-policy output QOS

verify operation with:

show xconnect all

show policy-map int gig1/1 output

[walliaralot]

Roman,

that's a goood white-page

thanks,

Kevin