ASR901: encapsulation default xconnect vs xconnect port mode

[emiliano.eusebio]

Hi to All,

I have a doubt about the possible difference between the xconnect with encapsulation default and the xconnect in port mode, is the same thing? Is one better than other? which is the correct implementation and configuration of the xconnect port mode?

Here is my configuration:

ASR901-3#sh run inter giga 0/5

Building configuration...

Current configuration : 236 bytes

!

interface GigabitEthernet0/5

no ip address

loopback driver

negotiation auto

no keepalive

service instance 1 ethernet

  encapsulation default

  xconnect 192.168.201.4 123 encapsulation mpls

!

end

Jan 10 17:52:47.338: AToM LDP[192.168.201.4]: Sending notif msg

Jan 10 17:52:47.338: AToM LDP[192.168.201.4]: Sending notif msg vc type 5, cbit 1,

Jan 10 17:52:47.338: AToM LDP[192.168.201.4]:   , vc id 123,

Jan 10 17:52:47.338: AToM LDP[192.168.201.4]:   group id 0, vc label 29,

Jan 10 17:52:47.338: AToM LDP[192.168.201.4]:   status 0x7/LDP 0x1/ADJ 0x0,

Jan 10 17:52:47.338: AToM LDP[192.168.201.4]:   mtu 9216, vc handle 0xF2000007

Jan 10 17:52:47.338: AToM LDP[192.168.201.4]:   VCCV CC 0x2, CV 0x2

ASR901-3#sh mpls l2 vc 123 detail

Local interface: Gi0/5 up, line protocol up, Ethernet:1 up

  Destination address: 192.168.201.4, VC ID: 123, VC status: up

    Output interface: Vl10, imposed label stack {20 25}

    Preferred path: not configured 

    Default path: active

    Next hop: 192.168.201.242

  Create time: 00:05:20, last status change time: 00:00:32

    Last label FSM state change time: 00:04:02

  Signaling protocol: LDP, peer 192.168.201.4:0 up

    Targeted Hello: 192.168.201.6(LDP Id) -> 192.168.201.4, LDP is UP

    Graceful restart: not configured and not enabled

    Non stop routing: not configured and not enabled

    Status TLV support (local/remote)   : enabled/supported

      LDP route watch                   : enabled

      Label/status state machine        : established, LruRru

      Last local dataplane   status rcvd: No fault

      Last BFD dataplane     status rcvd: Not sent

      Last BFD peer monitor  status rcvd: No fault

      Last local AC  circuit status rcvd: No fault

      Last local AC  circuit status sent: No fault

      Last local PW i/f circ status rcvd: No fault

      Last local LDP TLV     status sent: No fault

      Last remote LDP TLV    status rcvd: No fault

      Last remote LDP ADJ    status rcvd: No fault

    MPLS VC labels: local 29, remote 25

    Group ID: local 0, remote 0

    MTU: local 9216, remote 9216

    Remote interface description:

  Sequencing: receive disabled, send disabled

  Control Word: On (configured: autosense)

  Dataplane:

    SSM segment/switch IDs: 12300/8203 (used), PWID: 4

  VC statistics:

    transit packet totals: receive 0, send 0

    transit byte totals:   receive 0, send 0

    transit packet drops:  receive 0, seq error 0, send 0

ASR901-3#

the other side:

ASR901-1#sh run inter giga 0/6

Building configuration...

Current configuration : 147 bytes

!

interface GigabitEthernet0/6

no ip address

loopback driver

negotiation auto

no keepalive

xconnect 192.168.201.6 123 encapsulation mpls

end

Jan 11 18:49:01.934: AToM LDP[192.168.201.6]: Sending notif msg

Jan 11 18:49:01.934: AToM LDP[192.168.201.6]: Sending notif msg vc type 5, cbit 1,

Jan 11 18:49:01.934: AToM LDP[192.168.201.6]:   , vc id 123,

Jan 11 18:49:01.934: AToM LDP[192.168.201.6]:   group id 0, vc label 25,

Jan 11 18:49:01.934: AToM LDP[192.168.201.6]:   status 0x7/LDP 0x1/ADJ 0x0,

Jan 11 18:49:01.934: AToM LDP[192.168.201.6]:   mtu 9216, vc handle 0xAD000001

Jan 11 18:49:01.934: AToM LDP[192.168.201.6]:   VCCV CC 0x2, CV 0x2

ASR901-1#sh mpls l2 vc 123 det

Local interface: Gi0/6 up, line protocol up, Ethernet up

  Destination address: 192.168.201.6, VC ID: 123, VC status: up

    Output interface: Vl10, imposed label stack {18 29}

    Preferred path: not configured 

    Default path: active

    Next hop: 192.168.201.246

  Create time: 00:19:47, last status change time: 00:16:16

    Last label FSM state change time: 00:19:46

  Signaling protocol: LDP, peer 192.168.201.6:0 up

    Targeted Hello: 192.168.201.4(LDP Id) -> 192.168.201.6, LDP is UP

    Graceful restart: not configured and not enabled

    Non stop routing: not configured and not enabled

    Status TLV support (local/remote)   : enabled/supported

      LDP route watch                   : enabled

      Label/status state machine        : established, LruRru

      Last local dataplane   status rcvd: No fault

      Last BFD dataplane     status rcvd: Not sent

      Last BFD peer monitor  status rcvd: No fault

      Last local AC  circuit status rcvd: No fault

      Last local AC  circuit status sent: No fault

      Last local PW i/f circ status rcvd: No fault

      Last local LDP TLV     status sent: No fault

      Last remote LDP TLV    status rcvd: No fault

      Last remote LDP ADJ    status rcvd: No fault

    MPLS VC labels: local 25, remote 29

    Group ID: local 0, remote 0

    MTU: local 9216, remote 9216

    Remote interface description:

  Sequencing: receive disabled, send disabled

  Control Word: On (configured: autosense)

  Dataplane:

    SSM segment/switch IDs: 4097/4096 (used), PWID: 1

  VC statistics:

    transit packet totals: receive 0, send 0

    transit byte totals:   receive 0, send 0

    transit packet drops:  receive 0, seq error 0, send 0

ASR901-1#

In both cases the loopback driver is configured because I don't have these ports connected to anything. But this no matter at this point. The thing is that I have two ways to configure the same thing, but i don't know which is the correct.

Thanks a Lot

Emiliano

[Akash Agrawal · Accepted solution]

Hi Emiliano,

In both cases encapsulation for xconnect is mpls only. the other encapsulation tht you have defined under service-instance is to map an incoming tag to a service instance. If you are receiving a frame without any tag it will be mapped to service instance 1.

Say suppose instead of "encapsulation default" we confiure "encapsulation dot1q 10" then all frames coming with tag 10 will be mapped to this service instance.

Now major difference between both implementaiton is, one is port-based EoMPLS second is scalable EoMPLS. In port-based EoMPLS you can use this port only for one L2 circuit service, while in EVC framework you can create multiple service instance and use them for different services.

Probably the biggest advantage of the EVC framework is the ability to support multiple services per physical port. This means that under a single physical port you can have any of the following mixed together :

- 802.1q trunk

- 802.1q tunnel

- Local connect

- Scalable EoMPLS (EoMPLS xconnect)

- Multipoint Bridging (L2 bridging)

- Multipoint Bridging (VPLS, SVI-based EoMPLS)

- L3 termination

Regards,

Akash

[emiliano.eusebio]

Hi Akash,

Thanks for your clarification!

Best Regards

Emiliano

[marek.hajduczenia]

Hi,

I have a small problem with configuration similar to what was shown above. This is the configuration I was hoping to get to:

Router#show running-config interface gigabitEthernet 0/6
Building configuration...

Current configuration : 269 bytes
!
interface GigabitEthernet0/6
 description PW-TEST
 no ip address
 media-type auto-select
 no negotiation auto
 no keepalive
 service instance 2100 ethernet
  encapsulation dot1q 2100
  xconnect 97.76.237.4 12345 encapsulation mpls
   mtu 2000
 !
end

but every time I try to add a xconnect into an empty (shell) service instance 2100, I get the following error:

Router(config-if-srv)#xconnect 97.76.237.4 12345 encapsulation mpls
Port-based xconnect is not supported if EFP is configured, remove all EFP's and re-configure

The only way to add the xconnect instance is to add also

encapsulation default

into the service instance 2100 ethernet. However, that means that all untagged frames are mapped into this PW, and what I am trying to achieve is map only frames with dot1q vid 2100.

Attached is the complete config for the router in question. Any suggestions would be welcome.

Software version: ROM: System Bootstrap, Version 15.4(1r)S2, RELEASE SOFTWARE (fc1)

Thanks

Marek

[emiliano.eusebio]

Hi Marek,

I think that you need to delimit the EFP with an encapsulation, if you want to add a full port, Port-Based EoMPLS you must do it without an EFP:

!
interface GigabitEthernet0/11
 description CE_facing
 xconnect 111.0.1.1 501 encapsulation mpls
!

But if you need to delimit the service, you need to add an service instance attached to an encapsulation (default or with the vlan 2100)

The configuration must look like that you show in your description:

!
interface GigabitEthernet0/11
 description CE_facing
 service instance 2100 ethernet
  encapsulation dot1q 2100
  rewrite ingress tag pop 1 symmetric
  xconnect 111.0.1.1 501 encapsulation mpls
!

Remember that the encapsulation is the only way to identify the "traffic" inside of the interface.

Encapsulation

Encapsulation defines the matching criteria that maps a VLAN, a range of VLANs, Ethertype, or a combination of these to a service instance. Configure encapsulation in the service instance configuration mode. You must configure one encapsulation command per EFP (service instance)

Take a look to the following link:

http://www.cisco.com/c/en/us/td/docs/wireless/asr_901/Configuration/Guide/b_asr901-scg/b_asr901-scg_chapter_01111.html#con_1084502

and 

http://www.cisco.com/c/en/us/td/docs/wireless/asr_901/Configuration/Guide/b_asr901-scg/b_asr901-scg_chapter_01000.html

Best Regards

Emiliano

[marek.hajduczenia]

Thank you, Emiliano

I ended up building the service off Gi0/4 as follows

interface GigabitEthernet0/4
 no ip address
 load-interval 30
 negotiation auto
 no keepalive
 service instance 2100 ethernet
  encapsulation dot1q 2100
  rewrite ingress tag pop 1 symmetric
  xconnect 97.76.237.4 12345 encapsulation mpls
   mtu 2000
 !

but a Wireshark capture off the trunk interface to next network device shows that all transferred frames do not have any VLAN tag on them. There is MPLS header and pretty much that is it - I am sharing a preview of the packet captured on trunk interface off C901 and going into next NE. It does seem that once the frame matches outer VID 2100, it is stripped from the packet and it is then mapped into MPLS tunnel. That is not what I am trying to achieve. I'd like to keep the VLAN tag from customer side and just push it into the tunnel.

Marek

[emiliano.eusebio]

Hi Marek,

If you want to keep the original VLAN ID, so you must to use the following configuration;

!
interface GigabitEthernet0/6
 no ip address
 loopback driver
 negotiation auto
 no keepalive
 service instance 5 ethernet
  encapsulation dot1q 5
  rewrite ingress tag pop 1 symmetric
  xconnect 192.168.201.6 123 encapsulation mpls pw-class TEST
 !

pseudowire-class TEST
 encapsulation mpls
 interworking vlan

So with this you change the mode of the VC TYPE (default is type 5 (port-mode)) to 4 that is vlan mode

ASR901-1#sh mpls l2transport binding 
  Destination Address: 192.168.201.6,VC ID: 123
    Local Label:  25
        Cbit: 1,    VC Type: Eth VLAN,    GroupID: 0
        MTU: 9216,   Interface Desc: n/a
        VCCV: CC Type: RA [2]
              CV Type: LSPV [2]
    Remote Label: 27
        Cbit: 1,    VC Type: Eth VLAN,    GroupID: 0
        MTU: 9216,   Interface Desc: n/a
        VCCV: CC Type: RA [2]
              CV Type: LSPV [2]

Please try with this configuration and let me know if this works for you.

Best Regards

Emiliano