SlideShare a Scribd company logo
STATUS	
  Use	
  Case:	
  
Performance	
  Engineered	
  LSPs.	
  
Rob	
  Shakir,	
  E2E	
  Network	
  Architect,	
  BT.	
  
STATUS	
  BoF,	
  IETF	
  87.	
  
July	
  29,	
  2013.	
  
©	
  BriHsh	
  TelecommunicaHons	
  plc	
   Slide	
  2	
  
Key Requirement.
Create	
  LSPs	
  within	
  an	
  IP/MPLS	
  infrastructure	
  which:	
  	
  
•  Are	
  routed	
  away	
  from	
  the	
  SPT	
  based	
  on	
  performance	
  
constraints	
  (affinity,	
  latency,	
  SRLG	
  etc.).	
  
•  Or	
  based	
  on	
  coupling	
  with	
  other	
  LSPs	
  within	
  the	
  network	
  
(e.g.,	
  for	
  diversity	
  or	
  bidirecHonality).	
  
•  Provide	
  adequate	
  scale	
  to	
  support	
  per-­‐service	
  or	
  per-­‐flow	
  
constraints.	
  
•  Are	
  routed	
  according	
  to	
  distributed	
  CSPF	
  or	
  centrally	
  based	
  
on	
  service	
  requirements.	
  
©	
  BriHsh	
  TelecommunicaHons	
  plc	
   Slide	
  3	
  
Background – What’s the problem we’re trying to address?
•  In	
  IP/MPLS	
  networks,	
  we	
  have	
  a	
  concept	
  of	
  one	
  “base”	
  topology	
  –	
  which	
  is	
  the	
  SPT.	
  
•  One	
  set	
  of	
  logic	
  applied	
  to	
  choose	
  IGP	
  costs	
  –	
  used	
  to	
  route	
  all	
  services	
  within	
  this	
  topology.	
  
	
  
•  Problem	
  for	
  a	
  core	
  network	
  suppor:ng	
  mul:ple	
  services:	
  Not	
  all	
  services	
  have	
  the	
  same	
  logic	
  as	
  
to	
  the	
  constraints	
  for	
  their	
  rouHng	
  through	
  the	
  infrastructure.	
  
Co-­‐rou:ng	
  service	
  placement	
  based	
  on	
  consideraHon	
  of	
  
other	
  services	
  within	
  the	
  network.	
  
Pinned	
  paths	
  where	
  services	
  are	
  constrained	
  based	
  on	
  
underlying	
  path	
  resources.	
  
•  How	
  do	
  we	
  meet	
  the	
  requirement	
  for	
  such	
  constraints?	
  	
  
-  Transport	
  networks	
  have	
  generally	
  provided	
  such	
  constrained	
  paths.	
  
-  More	
  applicaHons	
  requiring	
  performance	
  guarantees.	
  
-  For	
  all	
  traffic	
  (e.g.,	
  Broadcast).	
  
-  A	
  subset	
  (e.g.,	
  voice	
  within	
  a	
  mulH-­‐service	
  VPN).	
  
Problem:	
  Provide	
  means	
  to	
  introduce	
  rouHng	
  constraints	
  which	
  diverge	
  from	
  the	
  SPT	
  on	
  a	
  per-­‐service	
  or	
  per-­‐
flow	
  basis,	
  uHlising	
  the	
  exisHng	
  underlying	
  IP/MPLS	
  network	
  infrastructure.	
  
©	
  BriHsh	
  TelecommunicaHons	
  plc	
   Slide	
  4	
  
Path Constraints and Technology Options.
•  Per-­‐service/flow	
  rou:ng	
  requires	
  a	
  significant	
  
increase	
  in	
  the	
  number	
  of	
  RSVP-­‐TE	
  LSPs	
  when	
  
compared	
  to	
  current	
  deployments:	
  
-  Number	
  of	
  LSPs	
  is	
  greater	
  than	
  full	
  mesh	
  (already	
  
not	
  recommended).	
  
-  Scale	
  limit	
  of	
  mid-­‐point	
  signalling	
  during	
  large	
  
failures.	
  
	
  
•  Limited	
  addi:onal	
  func:onality	
  is	
  offered	
  by	
  
having	
  mid-­‐point	
  state.	
  
-  Generally	
  only	
  admission	
  control.	
  
-  Required	
  in	
  a	
  subset	
  of	
  path	
  rouHng	
  
scenarios.	
  
Mid-­‐point	
  Overloading	
  –	
  Post-­‐Mortem	
  Model	
  
Unbounded	
  RSVP-­‐TE	
  queue	
  growth	
  based	
  on	
  
inability	
  to	
  process	
  PATH	
  messages	
  within	
  LSP	
  
retry	
  Hme	
  –	
  LSPs	
  never	
  successfully	
  re-­‐signal.	
  
•  Requirement	
  for	
  a	
  number	
  of	
  types	
  of	
  constrained	
  service/flow	
  rou:ng:	
  
-  Co-­‐rouHng.	
  
-  Considering	
  SRLG/Node/Link	
  diversity	
  or	
  bi-­‐direcHonal	
  paths.	
  
-  Affinity-­‐based	
  rouHng.	
  
-  Diverging	
  from	
  SPT	
  based	
  on	
  constraining	
  available	
  paths	
  by	
  colour/admin-­‐group.	
  
-  Performance-­‐managed	
  services.	
  
-  Latency,	
  available	
  bandwidth,	
  etc.	
  
•  Clearly,	
  a	
  number	
  of	
  these	
  constraints	
  can	
  be	
  delivered	
  by	
  RSVP-­‐TE	
  today.	
  
©	
  BriHsh	
  TelecommunicaHons	
  plc	
   Slide	
  5	
  
Suggested Architecture.
•  Use	
  segment	
  rou:ng	
  (STATUS	
  approach)	
  to	
  provide	
  means	
  to	
  instan:ate	
  the	
  data-­‐plane	
  paths.	
  
-  Removes	
  constraint	
  of	
  number	
  of	
  LSPs	
  that	
  can	
  be	
  created	
  by	
  removing	
  mid-­‐point	
  state.	
  
-  Stacked	
  labels	
  indicate	
  the	
  path	
  to	
  be	
  taken	
  through	
  the	
  network.	
  
-  Some	
  consideraHon	
  of	
  per-­‐label	
  semanHcs	
  may	
  be	
  required	
  (parHcularly	
  for	
  reversion).	
  
•  Re-­‐use	
  exis:ng	
  CSPF	
  machinery	
  where	
  it	
  is	
  applicable.	
  
-  Distributed	
  path	
  calculaHon	
  based	
  on	
  IGP	
  or	
  TED	
  influences	
  selecHon	
  (e.g.,	
  affinity	
  etc.)	
  
-  Extended	
  IGP	
  afributes	
  can	
  provide	
  increased	
  CSPF-­‐funcHonality.	
  
-  dra$-­‐previdi-­‐isis-­‐te-­‐metric-­‐extensions-­‐03	
  
-  dra$-­‐ie4-­‐ospf-­‐te-­‐metric-­‐extensions-­‐04	
  
	
  
•  Re-­‐use	
  exis:ng	
  PCE	
  to	
  provide	
  SID	
  stacks	
  where	
  global	
  visibility	
  is	
  required.	
  
-  Co-­‐computaHon	
  of	
  coupled	
  services	
  (bidirecHonal,	
  diverse	
  with	
  divergent	
  head-­‐ends).	
  
-  Stateful	
  PCE	
  can	
  provide	
  admission	
  control	
  where	
  required.	
  
-  Only	
  maintain	
  state	
  for	
  the	
  subset	
  of	
  services	
  where	
  it	
  is	
  required.	
  
	
  
•  Use	
  case	
  described	
  in	
  more	
  detail	
  in	
  dra$-­‐shakir-­‐rtgwg-­‐sr-­‐performance-­‐engineered-­‐lsps.	
  
	
  
•  Encourage	
  swig	
  progress	
  towards	
  standardised	
  soluHons	
  to	
  meet	
  these	
  requirements.	
  
-  Extend	
  IETF	
  technologies	
  to	
  meet	
  an	
  IP/MPLS	
  funcHonality	
  gap.	
  
-  ExisHng	
  WGs	
  provide	
  a	
  good	
  forum	
  with	
  experts	
  in	
  each	
  domain.	
  
©	
  BriHsh	
  TelecommunicaHons	
  plc	
   Slide	
  6	
  
Thanks!
Questions/Comments?

More Related Content

DOCX
IEEE 2014 JAVA NETWORKING PROJECTS Cost effective resource allocation of over...
PDF
From GMPLS to OpenFlow Control & Monitoring of Optical Networks
PPT
PDF
Coexistence of GMPLS and OpenFlow: rationale & approaches
PDF
Speed5G Workshop London presentation of the Speed5G RRM framework
PDF
ETE405-lec9.pdf
PDF
GMPLS (generalized mpls)
PPTX
WSN NETWORK -MAC PROTOCOLS - Low Duty Cycle Protocols And Wakeup Concepts – ...
IEEE 2014 JAVA NETWORKING PROJECTS Cost effective resource allocation of over...
From GMPLS to OpenFlow Control & Monitoring of Optical Networks
Coexistence of GMPLS and OpenFlow: rationale & approaches
Speed5G Workshop London presentation of the Speed5G RRM framework
ETE405-lec9.pdf
GMPLS (generalized mpls)
WSN NETWORK -MAC PROTOCOLS - Low Duty Cycle Protocols And Wakeup Concepts – ...

What's hot (19)

PDF
Load balancing and handoff in lte
PDF
A multi path routing algorithm for ip
DOCX
Cooperative load balancing and dynamic
PDF
Gmpls tutorial and_rand_e_network_implementation
PPT
PDF
The future potential of High Speed Uplink Packet Access in existing 3G netw...
PPT
Mac adhoc (1)
PDF
Cooperative load balancing and dynamic
PDF
Cooperative load balancing and dynamic channel allocation for cluster based m...
PDF
Cellular expert for lte planning 2
PDF
PPTX
2014 03-09, ofc workshop, ping pan
PDF
TRACK B: Multicores & Network On Chip Architectures/ Oren Hollander
PDF
Nokia IES Configuration guide
PDF
LTE Review - Load Balancing and Interfreq HO
PPTX
Classifications of wireless adhoc networks
PDF
Bonhomie
PPTX
Day two planning
PDF
PLNOG 8: Peter Ashwood-Smith - Shortest Path Bridging IEEE 802.1aq
Load balancing and handoff in lte
A multi path routing algorithm for ip
Cooperative load balancing and dynamic
Gmpls tutorial and_rand_e_network_implementation
The future potential of High Speed Uplink Packet Access in existing 3G netw...
Mac adhoc (1)
Cooperative load balancing and dynamic
Cooperative load balancing and dynamic channel allocation for cluster based m...
Cellular expert for lte planning 2
2014 03-09, ofc workshop, ping pan
TRACK B: Multicores & Network On Chip Architectures/ Oren Hollander
Nokia IES Configuration guide
LTE Review - Load Balancing and Interfreq HO
Classifications of wireless adhoc networks
Bonhomie
Day two planning
PLNOG 8: Peter Ashwood-Smith - Shortest Path Bridging IEEE 802.1aq
Ad

Similar to IETF87 - STATUS BoF: Performance Engineered LSPs (20)

DOCX
2014 IEEE JAVA NETWORKING COMPUTING PROJECT Cost effective resource allocatio...
DOCX
2014 IEEE JAVA NETWORKING PROJECT Cost effective resource allocation of overl...
PDF
2004 qof is_mpls_ospf
DOCX
JPJ1433 Cost-Effective Resource Allocation of Overlay Routing Relay Nodes
PPTX
MENOG-Segment Routing Introduction
PDF
Segment Routing: Prepare Your Network For New Business Models
DOCX
cost effective resource allocation of overlay routing relay nodes
PDF
PLNOG 13: Julian Lucek: Centralized Traffic Enginnering
PDF
Mikrotik load balansing
PDF
The Challenges of SDN/OpenFlow in an Operational and Large-scale Network
PDF
LREProxy module for Kamailio Presenation
PPT
Carrier Strategies for Backbone Traffic Engineering and QoS
PPTX
Multi-Protocol Label Switching
PPTX
5G RAN Slicing for Dublin Release.pptx
PPTX
TechWiseTV Workshop: Segment Routing for the Datacenter
PPT
Network Planning & Design: An Art or a Science?
PDF
SmartFlowwhitepaper
PPTX
FP7 PACE PCE Tutorial
DOCX
2014 IEEE JAVA NETWORKING PROJECT On the excess bandwidth allocation in isp t...
2014 IEEE JAVA NETWORKING COMPUTING PROJECT Cost effective resource allocatio...
2014 IEEE JAVA NETWORKING PROJECT Cost effective resource allocation of overl...
2004 qof is_mpls_ospf
JPJ1433 Cost-Effective Resource Allocation of Overlay Routing Relay Nodes
MENOG-Segment Routing Introduction
Segment Routing: Prepare Your Network For New Business Models
cost effective resource allocation of overlay routing relay nodes
PLNOG 13: Julian Lucek: Centralized Traffic Enginnering
Mikrotik load balansing
The Challenges of SDN/OpenFlow in an Operational and Large-scale Network
LREProxy module for Kamailio Presenation
Carrier Strategies for Backbone Traffic Engineering and QoS
Multi-Protocol Label Switching
5G RAN Slicing for Dublin Release.pptx
TechWiseTV Workshop: Segment Routing for the Datacenter
Network Planning & Design: An Art or a Science?
SmartFlowwhitepaper
FP7 PACE PCE Tutorial
2014 IEEE JAVA NETWORKING PROJECT On the excess bandwidth allocation in isp t...
Ad

More from Rob Shakir (8)

PDF
BGP OPERATIONAL Message
PDF
Reinforcing the Kitchen Sink - Aligning BGP-4 Error Handling with Modern Netw...
PDF
IETF80 - IDR/GROW BGP Error Handling Requirements
PDF
BGP Error Handling (NANOG 51)
PDF
BGP Error Handling - Developing an Operator-Led Approach in the IETF (UKNOF 18)
PDF
100GE in the Lab - LINX 71
PDF
LINX65 - Handling BGP Attribute Errors (Rob Shakir)
PDF
UKNOF16 - Enhancing BGP
BGP OPERATIONAL Message
Reinforcing the Kitchen Sink - Aligning BGP-4 Error Handling with Modern Netw...
IETF80 - IDR/GROW BGP Error Handling Requirements
BGP Error Handling (NANOG 51)
BGP Error Handling - Developing an Operator-Led Approach in the IETF (UKNOF 18)
100GE in the Lab - LINX 71
LINX65 - Handling BGP Attribute Errors (Rob Shakir)
UKNOF16 - Enhancing BGP

Recently uploaded (20)

PDF
A novel scalable deep ensemble learning framework for big data classification...
PDF
Unlocking AI with Model Context Protocol (MCP)
PDF
DP Operators-handbook-extract for the Mautical Institute
PDF
Getting Started with Data Integration: FME Form 101
PDF
DASA ADMISSION 2024_FirstRound_FirstRank_LastRank.pdf
PPTX
A Presentation on Touch Screen Technology
PDF
ENT215_Completing-a-large-scale-migration-and-modernization-with-AWS.pdf
PDF
project resource management chapter-09.pdf
PDF
From MVP to Full-Scale Product A Startup’s Software Journey.pdf
PPTX
OMC Textile Division Presentation 2021.pptx
PDF
Agricultural_Statistics_at_a_Glance_2022_0.pdf
PDF
Assigned Numbers - 2025 - Bluetooth® Document
PPTX
cloud_computing_Infrastucture_as_cloud_p
PPTX
Digital-Transformation-Roadmap-for-Companies.pptx
PDF
Enhancing emotion recognition model for a student engagement use case through...
PDF
Video forgery: An extensive analysis of inter-and intra-frame manipulation al...
PPTX
TLE Review Electricity (Electricity).pptx
PDF
Univ-Connecticut-ChatGPT-Presentaion.pdf
PDF
WOOl fibre morphology and structure.pdf for textiles
PPTX
A Presentation on Artificial Intelligence
A novel scalable deep ensemble learning framework for big data classification...
Unlocking AI with Model Context Protocol (MCP)
DP Operators-handbook-extract for the Mautical Institute
Getting Started with Data Integration: FME Form 101
DASA ADMISSION 2024_FirstRound_FirstRank_LastRank.pdf
A Presentation on Touch Screen Technology
ENT215_Completing-a-large-scale-migration-and-modernization-with-AWS.pdf
project resource management chapter-09.pdf
From MVP to Full-Scale Product A Startup’s Software Journey.pdf
OMC Textile Division Presentation 2021.pptx
Agricultural_Statistics_at_a_Glance_2022_0.pdf
Assigned Numbers - 2025 - Bluetooth® Document
cloud_computing_Infrastucture_as_cloud_p
Digital-Transformation-Roadmap-for-Companies.pptx
Enhancing emotion recognition model for a student engagement use case through...
Video forgery: An extensive analysis of inter-and intra-frame manipulation al...
TLE Review Electricity (Electricity).pptx
Univ-Connecticut-ChatGPT-Presentaion.pdf
WOOl fibre morphology and structure.pdf for textiles
A Presentation on Artificial Intelligence

IETF87 - STATUS BoF: Performance Engineered LSPs

  • 1. STATUS  Use  Case:   Performance  Engineered  LSPs.   Rob  Shakir,  E2E  Network  Architect,  BT.   STATUS  BoF,  IETF  87.   July  29,  2013.  
  • 2. ©  BriHsh  TelecommunicaHons  plc   Slide  2   Key Requirement. Create  LSPs  within  an  IP/MPLS  infrastructure  which:     •  Are  routed  away  from  the  SPT  based  on  performance   constraints  (affinity,  latency,  SRLG  etc.).   •  Or  based  on  coupling  with  other  LSPs  within  the  network   (e.g.,  for  diversity  or  bidirecHonality).   •  Provide  adequate  scale  to  support  per-­‐service  or  per-­‐flow   constraints.   •  Are  routed  according  to  distributed  CSPF  or  centrally  based   on  service  requirements.  
  • 3. ©  BriHsh  TelecommunicaHons  plc   Slide  3   Background – What’s the problem we’re trying to address? •  In  IP/MPLS  networks,  we  have  a  concept  of  one  “base”  topology  –  which  is  the  SPT.   •  One  set  of  logic  applied  to  choose  IGP  costs  –  used  to  route  all  services  within  this  topology.     •  Problem  for  a  core  network  suppor:ng  mul:ple  services:  Not  all  services  have  the  same  logic  as   to  the  constraints  for  their  rouHng  through  the  infrastructure.   Co-­‐rou:ng  service  placement  based  on  consideraHon  of   other  services  within  the  network.   Pinned  paths  where  services  are  constrained  based  on   underlying  path  resources.   •  How  do  we  meet  the  requirement  for  such  constraints?     -  Transport  networks  have  generally  provided  such  constrained  paths.   -  More  applicaHons  requiring  performance  guarantees.   -  For  all  traffic  (e.g.,  Broadcast).   -  A  subset  (e.g.,  voice  within  a  mulH-­‐service  VPN).   Problem:  Provide  means  to  introduce  rouHng  constraints  which  diverge  from  the  SPT  on  a  per-­‐service  or  per-­‐ flow  basis,  uHlising  the  exisHng  underlying  IP/MPLS  network  infrastructure.  
  • 4. ©  BriHsh  TelecommunicaHons  plc   Slide  4   Path Constraints and Technology Options. •  Per-­‐service/flow  rou:ng  requires  a  significant   increase  in  the  number  of  RSVP-­‐TE  LSPs  when   compared  to  current  deployments:   -  Number  of  LSPs  is  greater  than  full  mesh  (already   not  recommended).   -  Scale  limit  of  mid-­‐point  signalling  during  large   failures.     •  Limited  addi:onal  func:onality  is  offered  by   having  mid-­‐point  state.   -  Generally  only  admission  control.   -  Required  in  a  subset  of  path  rouHng   scenarios.   Mid-­‐point  Overloading  –  Post-­‐Mortem  Model   Unbounded  RSVP-­‐TE  queue  growth  based  on   inability  to  process  PATH  messages  within  LSP   retry  Hme  –  LSPs  never  successfully  re-­‐signal.   •  Requirement  for  a  number  of  types  of  constrained  service/flow  rou:ng:   -  Co-­‐rouHng.   -  Considering  SRLG/Node/Link  diversity  or  bi-­‐direcHonal  paths.   -  Affinity-­‐based  rouHng.   -  Diverging  from  SPT  based  on  constraining  available  paths  by  colour/admin-­‐group.   -  Performance-­‐managed  services.   -  Latency,  available  bandwidth,  etc.   •  Clearly,  a  number  of  these  constraints  can  be  delivered  by  RSVP-­‐TE  today.  
  • 5. ©  BriHsh  TelecommunicaHons  plc   Slide  5   Suggested Architecture. •  Use  segment  rou:ng  (STATUS  approach)  to  provide  means  to  instan:ate  the  data-­‐plane  paths.   -  Removes  constraint  of  number  of  LSPs  that  can  be  created  by  removing  mid-­‐point  state.   -  Stacked  labels  indicate  the  path  to  be  taken  through  the  network.   -  Some  consideraHon  of  per-­‐label  semanHcs  may  be  required  (parHcularly  for  reversion).   •  Re-­‐use  exis:ng  CSPF  machinery  where  it  is  applicable.   -  Distributed  path  calculaHon  based  on  IGP  or  TED  influences  selecHon  (e.g.,  affinity  etc.)   -  Extended  IGP  afributes  can  provide  increased  CSPF-­‐funcHonality.   -  dra$-­‐previdi-­‐isis-­‐te-­‐metric-­‐extensions-­‐03   -  dra$-­‐ie4-­‐ospf-­‐te-­‐metric-­‐extensions-­‐04     •  Re-­‐use  exis:ng  PCE  to  provide  SID  stacks  where  global  visibility  is  required.   -  Co-­‐computaHon  of  coupled  services  (bidirecHonal,  diverse  with  divergent  head-­‐ends).   -  Stateful  PCE  can  provide  admission  control  where  required.   -  Only  maintain  state  for  the  subset  of  services  where  it  is  required.     •  Use  case  described  in  more  detail  in  dra$-­‐shakir-­‐rtgwg-­‐sr-­‐performance-­‐engineered-­‐lsps.     •  Encourage  swig  progress  towards  standardised  soluHons  to  meet  these  requirements.   -  Extend  IETF  technologies  to  meet  an  IP/MPLS  funcHonality  gap.   -  ExisHng  WGs  provide  a  good  forum  with  experts  in  each  domain.  
  • 6. ©  BriHsh  TelecommunicaHons  plc   Slide  6   Thanks! Questions/Comments?