PBB TE
PBB-TE



 (c) Anuradha Udunuwara   1
Agenda

• Introduction
• QinQ
• PBB
• PBB-TE
• Conclusion
             (c) Anuradha Udunuwara   2
Introduction




   (c) Anuradha Udunuwara   3
Connection-less (CL) networking based
                ( L)    w      g
on Ethernet and IP/MPLS has been
around for some time. There are m y
       f     m    m .            many
carriers who have not only implemented
these technologies, but have used them
               g ,                   m
to completely replace their legacy
network So, why are we discussing
network. So
connection-oriented(CO) packet
networking?
  t   ki ?

               (c) Anuradha Udunuwara   4
Carrier Ethernet options
• PBB-TE MPLS TP
  PBB TE, MPLS-TP
 – no control plane (like MPLS in IP/MPLS)
 – uses EMS/NMS for provisioning
   • NMS based control plane
   • centralized intelligence on a server
   • vendor sticky
 – no multi point
 – smaller forwarding tables
• MPLS
                     (c) Anuradha Udunuwara   5
QinQ




(c) Anuradha Udunuwara   6
QinQ
= 802 1QiQ
  802.1QiQ
= 802.1ad
= P id Bridging (PB)
  Provider B id i

[C-VLAN inner, S-VLAN outer]
In this scenario provider needs to learn
         scenario,
  all customer MACs.

                  (c) Anuradha Udunuwara   7
Where is it most appropriate to use Q-in-Q and
  MAC-in-MAC i
  MAC i MAC in a Hi Hierarchical L2 Network?
                            hi l    N t    k?

• MAC-in-MAC core domain surrounded by
  MAC in MAC
  Q-in-Q access domains.
  – Removes need for global Q-tag management
     • Locally convert MAC-in-MAC service tag to local
       Q g
       Q-tag
  – Reduces number of MAC addresses seen by
    MAC-in-MAC domain to number of Q-in-Q
    domains.


                      (c) Anuradha Udunuwara         8
Why Not VPLS End-to-End?
               End to End?
• VPLS has scalability issues
  – E.g. to connect 5 PE devices requires 20
    LSPs [(n*(n-1)/2)*2] 40 PEs: 1 536
          [(n (n 1)/2) 2],        1,536.
  – Large bandwidth loss due to broadcast
    retransmissions.
• VPLS requires new features
  – Protection OAM discovery
    Protection, OAM,


                    (c) Anuradha Udunuwara     9
PBB




(c) Anuradha Udunuwara   10
PBB
= 802.1ah
  802 1ah
= Provider Backbone Bridging(PBB)
= MAC in MAC
       i




                 (c) Anuradha Udunuwara   11
PBB
• PBB uses STP (PBB-TE does not use
  STP)
• PBB stops the access MAC
             h
  complexity form core.
     p     y
  – Ex:- run PBB at access and VPLS in
    the core (so the VSI needs to learn
    only the S-MAC not 100,000 of C-
    MACs)
                 (c) Anuradha Udunuwara   12
PBB TE
PBB-TE




 (c) Anuradha Udunuwara   13
PBB TE
               PBB-TE

= 802 1Qay
  802.1Qay
= Provider Backbone Bridging –Traffic
  Engineering
  E i      i
= PBT (Provider Backbone Transport) by
                               p      y
  Nortel

Both PBB-TE and MPLS-TP has “SDH-like”
 operation
                 (c) Anuradha Udunuwara   14
PBB-TE builds on Ethernet
• MPLS-TP requires an underlying L1/L2
  p
  protocol …… at a time when Ethernet is
  migrating to the MAN and WAN as the
  ubiquitous underlying L1/L2 protocol
• MPLS TP proposes t use a L2 switching l b l
  MPLS-TP             to            it hi label
  that is unrelated to Ethernet’s addressing
  scheme
  – And define its own control plane and management
    plane
• PBB TE proposes t use a Eth
  PBB-TE          to      Ethernet’s
                                 t’
  addressing scheme
  – And leverage Ethernet’s PBB provider header,
                 Ethernet s              header
    management plane and control plane
                      (c) Anuradha Udunuwara          15
• PBB-TE was designed to handle point-
  PBB TE                          point
  to-point data services in a PBB network.




                  (c) Anuradha Udunuwara     16
While PBB provides for Carrier-Ethernet
                  switching,
    PBB-TE provides for Carrier-Ethernet
                  Transport
     through the addition of the following
               characteristics:

• Ability to provision protected virtual circuits
• Improved control of the transport layer's
                                      layer s
  operational state through SDH-like OA&M
• Improved network and survivability
  mechanisms such as protection and
  restoration as seen in SDH
                     (c) Anuradha Udunuwara         17
The PBB-TE Advantage

• Secure Demarcation
   – Customer networks may have overlapping addresses, which
     could cause forwarding conflicts in the Telecom Operators
     network.


• Customer Separation & Service Scalability
   – PBB provides 16 million unique service id
              id       illi    i        i identifiers, apart from
                                               ifi           f
     providing 2 transport tunnels.


• Single-ended Provisioning
   – only the end-nodes have to be provisioned


                           (c) Anuradha Udunuwara                18
PBB/PBB-TE combination

• PBB and PBB-TE can be deployed in the same network,
                             p y                         ,
  and can also share a fiber pair.
• Network nodes use the B-VLAN field in the PBB
  header t diff
  h d to differentiate between th PBB and PBB TE
                     ti t b t       the        d PBB-TE
  traffic.
• While PBB traff c can be protected using ERPS (ITU-
  Wh le BB traffic                     us ng E S (
  T G.8032), the PBB-TE traffic is protected using EPS
  (ITU-T G.8031).
• H
  Hence, while PBB-TE can be used to transport
           hil PBB TE       b      dt t         t
  services that require traffic-engineering, PBB can be
  used for any-to-any or multipoint services, all on the
              y      y         p
  same network.
                        (c) Anuradha Udunuwara           19
Conclusion




  (c) Anuradha Udunuwara   20
• All three (MPLS/VPLS, MPLS-TP, PBB-TE)
  solve the same problem
• MPLS-TP is born out of realization that
  M LS/V LS s
  MPLS/VPLS is complex and expensive
                             expens ve
  – Essentially confirms providers only really have a
    long term choice between MPLS-TP and PBB-TE
• So why chose PBB-TE over MPLS-TP, or vice-
  versa?



                       (c) Anuradha Udunuwara           21

PBB-TE

  • 1.
    PBB TE PBB-TE (c)Anuradha Udunuwara 1
  • 2.
    Agenda • Introduction • QinQ •PBB • PBB-TE • Conclusion (c) Anuradha Udunuwara 2
  • 3.
    Introduction (c) Anuradha Udunuwara 3
  • 4.
    Connection-less (CL) networkingbased ( L) w g on Ethernet and IP/MPLS has been around for some time. There are m y f m m . many carriers who have not only implemented these technologies, but have used them g , m to completely replace their legacy network So, why are we discussing network. So connection-oriented(CO) packet networking? t ki ? (c) Anuradha Udunuwara 4
  • 5.
    Carrier Ethernet options •PBB-TE MPLS TP PBB TE, MPLS-TP – no control plane (like MPLS in IP/MPLS) – uses EMS/NMS for provisioning • NMS based control plane • centralized intelligence on a server • vendor sticky – no multi point – smaller forwarding tables • MPLS (c) Anuradha Udunuwara 5
  • 6.
  • 7.
    QinQ = 802 1QiQ 802.1QiQ = 802.1ad = P id Bridging (PB) Provider B id i [C-VLAN inner, S-VLAN outer] In this scenario provider needs to learn scenario, all customer MACs. (c) Anuradha Udunuwara 7
  • 8.
    Where is itmost appropriate to use Q-in-Q and MAC-in-MAC i MAC i MAC in a Hi Hierarchical L2 Network? hi l N t k? • MAC-in-MAC core domain surrounded by MAC in MAC Q-in-Q access domains. – Removes need for global Q-tag management • Locally convert MAC-in-MAC service tag to local Q g Q-tag – Reduces number of MAC addresses seen by MAC-in-MAC domain to number of Q-in-Q domains. (c) Anuradha Udunuwara 8
  • 9.
    Why Not VPLSEnd-to-End? End to End? • VPLS has scalability issues – E.g. to connect 5 PE devices requires 20 LSPs [(n*(n-1)/2)*2] 40 PEs: 1 536 [(n (n 1)/2) 2], 1,536. – Large bandwidth loss due to broadcast retransmissions. • VPLS requires new features – Protection OAM discovery Protection, OAM, (c) Anuradha Udunuwara 9
  • 10.
  • 11.
    PBB = 802.1ah 802 1ah = Provider Backbone Bridging(PBB) = MAC in MAC i (c) Anuradha Udunuwara 11
  • 12.
    PBB • PBB usesSTP (PBB-TE does not use STP) • PBB stops the access MAC h complexity form core. p y – Ex:- run PBB at access and VPLS in the core (so the VSI needs to learn only the S-MAC not 100,000 of C- MACs) (c) Anuradha Udunuwara 12
  • 13.
    PBB TE PBB-TE (c)Anuradha Udunuwara 13
  • 14.
    PBB TE PBB-TE = 802 1Qay 802.1Qay = Provider Backbone Bridging –Traffic Engineering E i i = PBT (Provider Backbone Transport) by p y Nortel Both PBB-TE and MPLS-TP has “SDH-like” operation (c) Anuradha Udunuwara 14
  • 15.
    PBB-TE builds onEthernet • MPLS-TP requires an underlying L1/L2 p protocol …… at a time when Ethernet is migrating to the MAN and WAN as the ubiquitous underlying L1/L2 protocol • MPLS TP proposes t use a L2 switching l b l MPLS-TP to it hi label that is unrelated to Ethernet’s addressing scheme – And define its own control plane and management plane • PBB TE proposes t use a Eth PBB-TE to Ethernet’s t’ addressing scheme – And leverage Ethernet’s PBB provider header, Ethernet s header management plane and control plane (c) Anuradha Udunuwara 15
  • 16.
    • PBB-TE wasdesigned to handle point- PBB TE point to-point data services in a PBB network. (c) Anuradha Udunuwara 16
  • 17.
    While PBB providesfor Carrier-Ethernet switching, PBB-TE provides for Carrier-Ethernet Transport through the addition of the following characteristics: • Ability to provision protected virtual circuits • Improved control of the transport layer's layer s operational state through SDH-like OA&M • Improved network and survivability mechanisms such as protection and restoration as seen in SDH (c) Anuradha Udunuwara 17
  • 18.
    The PBB-TE Advantage •Secure Demarcation – Customer networks may have overlapping addresses, which could cause forwarding conflicts in the Telecom Operators network. • Customer Separation & Service Scalability – PBB provides 16 million unique service id id illi i i identifiers, apart from ifi f providing 2 transport tunnels. • Single-ended Provisioning – only the end-nodes have to be provisioned (c) Anuradha Udunuwara 18
  • 19.
    PBB/PBB-TE combination • PBBand PBB-TE can be deployed in the same network, p y , and can also share a fiber pair. • Network nodes use the B-VLAN field in the PBB header t diff h d to differentiate between th PBB and PBB TE ti t b t the d PBB-TE traffic. • While PBB traff c can be protected using ERPS (ITU- Wh le BB traffic us ng E S ( T G.8032), the PBB-TE traffic is protected using EPS (ITU-T G.8031). • H Hence, while PBB-TE can be used to transport hil PBB TE b dt t t services that require traffic-engineering, PBB can be used for any-to-any or multipoint services, all on the y y p same network. (c) Anuradha Udunuwara 19
  • 20.
    Conclusion (c)Anuradha Udunuwara 20
  • 21.
    • All three(MPLS/VPLS, MPLS-TP, PBB-TE) solve the same problem • MPLS-TP is born out of realization that M LS/V LS s MPLS/VPLS is complex and expensive expens ve – Essentially confirms providers only really have a long term choice between MPLS-TP and PBB-TE • So why chose PBB-TE over MPLS-TP, or vice- versa? (c) Anuradha Udunuwara 21