• Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Be the first to comment
    Be the first to like this
No Downloads

Views

Total Views
2,112
On Slideshare
0
From Embeds
0
Number of Embeds
0

Actions

Shares
Downloads
115
Comments
0
Likes
0

Embeds 0

No embeds

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
    No notes for slide

Transcript

  • 1. High Availability for IPTV Søren Andreasen CCIE #3252 Cisco DK © 2005 Cisco Systems, Inc. All rights reserved. Cisco Confidential 1
  • 2. Agenda • IPTV Architecture overview • High Availability (HA) in the core and edge – Router Level HA with SSO – Protocol HA with NSF/GR (IGP, BGP, Multicast) – MPLS HA (LDP NSF/GR, BGP + label NSF, FRR, PW Redundancy) – Fast Convergence (Multicast FC, Fast channel change, LDP-IGP sync, LDP session protection) • Video Source Redundancy • Conclusion © 2005 Cisco Systems, Inc. All rights reserved. 2
  • 3. Tripleplay world Portal Broadband Subscriber Policy Billing Identity Policy Manager Database Definition Policy and Service Management Layer Residential Ethernet Access IP / MPLS / Ethernet Ethernet IP / MPLS Core ISG/BRAS Aggregation ETTx RGW Si Distributed: Centralized: L2 PW, VPLS H-VPLS Cable L3 VPN L3 VPN DSL Local CO PE-R / PE-S Metro CO STB PE-R CORE STB PON Service Control Engine Hosted Business Apps iFrame Cache (Storage, Centrex, Security, Gaming) VoD VoIP Video Broadcast Service/Application Layer © 2005 Cisco Systems, Inc. All rights reserved. 3
  • 4. Agenda • IPTV Architecture overview • High Availability (HA) in the core and edge – Router Level HA with SSO – Protocol HA with NSF/GR (IGP, BGP, Multicast) – MPLS HA (LDP NSF/GR, BGP + label NSF, FRR, PW Redundancy) – Fast Convergence (Multicast FC, Fast channel change, LDP-IGP sync, LDP session protection) • Video Source Redundancy • Conclusion © 2005 Cisco Systems, Inc. All rights reserved. 4
  • 5. High Availability • End-to-end protection against network failures • Preserve end-to-end network service connectivity • Isolate control plane failures from forwarding plane • Separation of control and forwarding plane should allow forwarding to continue while control plane recovers (NSF) © 2005 Cisco Systems, Inc. All rights reserved. 5
  • 6. HA Mechanisms • Component and Device Level Resiliency – Hardware and software component resiliency • Distributed line cards, route processors, Modular operating software – Stateful Switch-Over between RPs (SSO) – Control/forwarding plane decoupling; Non-Stop Forwarding (NSF) – Graceful Restart of protocols (GR) • Network Level Resiliency – Optimized convergence algorithms, speeding up network recovery © 2005 Cisco Systems, Inc. All rights reserved. 6
  • 7. Device Level HA: NSF With SSO • Stateful Switch-Over (SSO): zero interruption to protocol sessions Route Processors – Active RP synchronizes information with standby RP – Session state maintained for high RP1 RP2 availability-aware protocols on ACTIVE STANDBY standby RP – Standby RP takes control when active RP is compromised Bus • Non-Stop Forwarding (NSF): minimal or no packet loss – Packet forwarding continues during reestablishment of peering relationships Line Cards – No route flaps between participating neighbor routers © 2005 Cisco Systems, Inc. All rights reserved. 7
  • 8. SSO in action HA-Router(config)#redundancy HA-Router#show redundancy Active GRP in slot 0: HA-Router(config-red)#mode ? Standby GRP in slot 9: rpr Route Processor Redundancy rpr-plus Route Processor Redundancy Plus Preferred GRP: none sso Stateful Switchover (Hot Operating Redundancy Mode: SSO Standby) Auto synch: startup-config running- config HA-Router(config-red)#mode sso switchover timer 3 seconds [default] © 2005 Cisco Systems, Inc. All rights reserved. 8
  • 9. Agenda • IPTV Architecture overview • High Availability (HA) in the core and edge – Router Level HA with SSO – Protocol HA with NSF/GR (IGP, BGP, Multicast) – MPLS HA (LDP NSF/GR, BGP + label NSF, FRR, PW Redundancy) – Fast Convergence (Multicast FC, Fast channel change, LDP-IGP sync, LDP session protection) • Video Source Redundancy • Conclusion © 2005 Cisco Systems, Inc. All rights reserved. 9
  • 10. GR/NSF Fundamentals • NonStop Forwarding (NSF) is a way to continue forwarding packets while the control plane is recovering from a failure. • Graceful Restart (GR) is a way to rebuild forwarding information in routing protocols when the control plane has recovered from a failure. © 2005 Cisco Systems, Inc. All rights reserved. 10
  • 11. GR/NSF Fundamentals (uden GR/NSF) • Router A loses its control plane for some period of Control Data A time. • It will take some time for Router B to recognize this failure, and react to it. Control Data B © 2005 Cisco Systems, Inc. All rights reserved. 11
  • 12. GR/NSF Fundamentals (uden GR/NSF) reset • During the time that A has failed, and B has not detected the failure, B will continue Control Data A forwarding traffic through A. • Once the control plane resets, the data plane will reset as well, and this traffic will be dropped. • NSF reduces or eliminates the traffic dropped while A’s control plane is down. Control Data B © 2005 Cisco Systems, Inc. All rights reserved. 12
  • 13. GR/NSF Fundamentals no reset • If A is NSF capable, the control plane will not reset the data plane when it restarts Control Data A • Instead, the forwarding information in the data plane is marked as stale. • Any traffic B sends to A will still be switched based on the last known forwarding information. Control Data B mark forwarding information as stale © 2005 Cisco Systems, Inc. All rights reserved. 13
  • 14. GR/NSF Fundamentals • While A’s control plane is down, the routing protocol hold timer on B counts Control Data A down.... • A has to come back up and signal B before B’s hold timer expires, or B will route around it. • When A comes back up, it signals B that it is still forwarding traffic, and would like to resync. Control Data B Hold Timer: 15 6 7 8 9 10 11 12 13 14 © 2005 Cisco Systems, Inc. All rights reserved. 14
  • 15. GR/NSF Design Goals • No Link Flap • CEF table sync to the standby RP • LC CEF Table not cleared on switchover • Packet forwarding during switchover while routing is converging on Standby RP • Maintain peer relationship, no adjacency flapping with peers • Limit restart to be local event, not network wide Ultimate Goal: Achieve 0% Packet Loss © 2005 Cisco Systems, Inc. All rights reserved. 15
  • 16. IS-IS GR/NSF • Use the nsf command under router isis the router isis configuration nsf A mode to enable graceful .... restart. • Show isis nsf can be used to verify graceful restart is operational. router isis nsf .... router#show isis nsf NSF is ENABLED, mode 'cisco' B RP is ACTIVE, standby ready, bulk sync complete NSF interval timer expired (NSF restart enabled) Checkpointing enabled, no errors Local state:ACTIVE, Peer state:STANDBY HOT, Mode:SSO © 2005 Cisco Systems, Inc. All rights reserved. 16
  • 17. OSPF GR/NSF • Use the nsf command under router ospf 100 the router ospf configuration nsf A mode to enable graceful .... restart. • Show ip ospf can be used to verify graceful restart is operational. router ospf 100 nsf .... router#sh ip ospf Routing Process "ospf 100" with ID 10.1.1.1 .... B Non-Stop Forwarding enabled, last NSF restart 00:02:06 ago (took 44 secs) router#show ip ospf neighbor detail Neighbor 3.3.3.3, interface address 170.10.10.3 .... Options is 0x52 LLS Options is 0x1 (LR), last OOB-Resync 00:02:22 ago © 2005 Cisco Systems, Inc. All rights reserved. 17
  • 18. BGP GR/NSF • Use the bgp graceful-restart command under the router bgp router bgp 65000 configuration mode to enable bgp graceful-restart A graceful restart. .... • Show ip bgp neighbors can be used to verify graceful restart is operational. router bgp 65501 bgp graceful-restart .... router#show ip bgp neighbors x.x.x.x .... B Neighbor capabilities: .... Graceful Restart Capabilty:advertised and received Remote Restart timer is 120 seconds Address families preserved by peer: IPv4 Unicast, IPv4 Multicast © 2005 Cisco Systems, Inc. All rights reserved. 18
  • 19. Multicast HA : PIM/SSM in the Core How Triggered PIM Join(s) work when Active Route Processor Fails: Periodic PIM Joins 1 Active Route Processor receives periodic “Special” PIM Hello PIM Joins in steady-state Triggered PIM Joins STANDBY ACTIVE ACTIVE 2 Active Route Processor fails Failure 3 Standby Route Processor takes over Line Card Line Card Line Card Line Card 4 “Special” PIM Hello is sent out 5 Triggers adjacent PIM neighbors to resend PIM Joins refreshing state of distribution tree(s) preventing them from timing out © 2005 Cisco Systems, Inc. All rights reserved. 19
  • 20. Multicast HA: IGMP at the edge Provides Seamless Forwarding of Multicast Traffic During Supervisor Switchovers How it works: STANDBY ACTIVE ACTIVE 1 Global sync of IGMP Data Structures Failure occurs when standby Supervisor comes online 2 Periodic syncs are performed as changes occur 3 Active Supervisor fails global 4 Standby Supervisor takes over sync IGMP IGMP 5 Synced IGMP Data Structures are utilized Data Data to provide Layer-2 Non-Stop Forwarding (NSF) Structures periodic Structures of multicast traffic syncs © 2005 Cisco Systems, Inc. All rights reserved. 20
  • 21. Agenda • IPTV Architecture overview • High Availability (HA) in the core and edge – Router Level HA with SSO – Protocol HA with NSF/GR (IGP, BGP, Multicast) – MPLS HA (LDP NSF/GR, BGP + label NSF, FRR, PW Redundancy) – Fast Convergence (Multicast FC, Fast channel change, LDP-IGP sync, LDP session protection) • Video Source Redundancy • Conclusion © 2005 Cisco Systems, Inc. All rights reserved. 21
  • 22. MPLS HA – Component Resiliency • MPLS HA features extend NSF with SSO for: – LDP (for L3VPNs and L2VPNs) – BGP (L3VPNs) – RSVP (for TE) • Minimal disruption to MPLS forwarding plane due to route processor control plane failures – Includes MPLS control plane failures (LDP, BGP, RSVP) © 2005 Cisco Systems, Inc. All rights reserved. 22
  • 23. MPLS HA – Network Resiliency • MPLS control plane protocol enhancements to improve failure detection time and network convergence • Graceful Restart (GR) – LDP, BGP, RSVP • Fast Convergence – LDP (IGP sync), LDP session protection • TE FRR – Link protection – Node protection • Pseudowire Redundancy © 2005 Cisco Systems, Inc. All rights reserved. 23
  • 24. MPLS Graceful Restart + NSF/SSO No MPLS HA support 1 Control Plane Control Plane Control Plane 2 2 Data Plane 3 Data Plane 3 3 Data Plane 3 LSPs LSPs PE P PE MPLS HA Support: NSF/SSO + Graceful Restart 1 Control Plane Control Plane Control Plane 2 2 Data Plane Data Plane Data Plane LSPs LSPs PE P PE © 2005 Cisco Systems, Inc. All rights reserved. 24
  • 25. LDP NSF/GR • LDP NSF/SSO/GR is also known as LDP NSF • LDP NSF allows a route processor to recover from disruption in LDP control plane without losing its MPLS forwarding state • LDP NSF works with LDP sessions between directly connected peers as well as with targeted sessions • LDP HA Key Elements – Checkpointing local label bindings (synkronisering mellem RP’s) • On devices with route processor redundancy – LDP graceful restart capability • On participating PEs, RRs, and P routers © 2005 Cisco Systems, Inc. All rights reserved. 25
  • 26. LDP NSF/GR Key Elements: Checkpointing Local Label Bindings • The checkpointing function is enabled Local Label Bindings by default RP A RP B (Active) (Standby) • The checkpointing function copies active RP’s LDP local label bindings to IPC the backup RP • For the first round, all the labels are copied from active to back up RP RP Checkpointing RP • Periodic incremental updates are done to reflect new routes that have been learned or routes that have been removed and/or when labels are allocated or freed • Label bindings on backup RP are marked checkpointed Line Card Line Card Line Card • This marking is removed when it becomes active RP © 2005 Cisco Systems, Inc. All rights reserved. 26
  • 27. LDP NSF/GR Operation Rb(config)#mpls ldp LDP GR: 20.0.0.0/8:: updating LDP GR: 10.0.0.0/8:: updating graceful-restart binding from 1.1.1.1:0, inst 1:: binding from 1.1.1.1:0, inst 1:: marking stale; LDP Adj LDP Adj marking stale; UP UP Primary A Standby C LDP B LDP Reset Reset LDP GR: 20.0.0.0//8:: LDP GR: 10.0.0.0//8:: refreshing stale binding from LDP refreshing stale binding from 1.1.1.1:0, inst 1 -> inst 2 Restart 1.1.1.1:0, inst 1 -> inst 2 • LDP paths established, LDP GR negotiated • When RP fails on LSRb, communication between peers is lost; LSRb encounters a LDP restart, while LSRa and LSRc encounter an LDP session reset • LSRa and LSRc mark all the label bindings from LSRb as stale, but continue to use the same bindings for MPLS forwarding • LSRa and LSRc attempt to reestablish an LDP session with Rb • LSRb restarts and marks all of its forwarding entries as stale; at this point, LDP state is in restarting mode • LSRa and LSRc reestablish LDP sessions with Rb, but keep their stale label bindings; at this point, the LDP state is in recovery mode • All routers re-advertise their label binding info; stale flags are removed if a label has been relearned; new LFIB table is ready © 2005 Cisco Systems, Inc. All rights reserved. 27
  • 28. BGP NSF/GR for Inter-AS and CSC Enable MPLS Nonstop Forwarding on an Interface that Uses BGP as the Label Distribution Protocol (Inter-as and CsC Environment) mpls bgp forwarding ASBR-1 ASBR-2 AS #100 AS #200 eBGP IPv4 + AS1_PE1 Labels AS2_PE1 mpls bgp forwarding mpls bgp forwarding © 2005 Cisco Systems, Inc. All rights reserved. 28
  • 29. MPLS TE Fast Re-Route (FRR) Link Protection • IP routing protocols (e.g., R3 OSPF, BGP) may be tuned to convergence within a few seconds • Some traffic (e.g. voice) will R1 R2 R4 R5 require more aggressive convergence time Node Protection – Typically 50 ms or less R3 R4 R5 • MPLS TE FRR offers protection against network failures R1 R2 R6 R7 R8 – Link protection R9 – Node protection Protected Path Backup Path © 2005 Cisco Systems, Inc. All rights reserved. 29
  • 30. FRR Link Protection R3 • Creation of Next-hop Backup Tunnel – Parallel path around 20 20 11 11 11 11 protected link to next hop HE PLR • On link failure detection Point of Local Repair (PLR) R1 10 10 R2 11 11 R4 R6 swaps label and pushes backup label – Traffic sent over backup R5 path • PLR notifies TE Head End x x Label for the backup LSP (HE), which triggers global TE path re-optimization x x Label for the protected LSP Protected Path Backup Path © 2005 Cisco Systems, Inc. All rights reserved. 30
  • 31. FRR Node Protection • Creation of Next-next-hop 21 21 12 12 Backup Tunnel – Parallel path around R3 R4 20 20 12 12 R5 protected node to next- 12 12 next-hop HE PLR R6 • Node failure triggers Point 11 12 R1 10 R2 11 12 R7 R8 of Local Repair (PLR) to 10 swap label and push backup label R9 – Traffic sent over backup path around failed node x x Label for the backup LSP • PLR notifies TE Head End (HE), which triggers global x x Label for the protected LSP TE path re-optimization Protected Path Backup Path © 2005 Cisco Systems, Inc. All rights reserved. 31
  • 32. FRR in triple play with P2MP-TE Basic RSVP-TE P2MP Tunnel Setup Distribution/ Source Service Edge Core Access Receiver Multicast Packet Primary RSVP Tunnel Multicast Packet + MPLS Label P P Layer 2 Layer 2 Switch Switch PE MPLS Core CE PE PE CE Source Receiver P P Backup RSVP Tunnel © 2005 Cisco Systems, Inc. All rights reserved. 32
  • 33. Multiple Points of Failure Still Exist Distribution/ Source Service Edge Core Access Receiver Primary RSVP Tunnel P P Layer 2 Layer 2 Switch Switch X X XX X X PE MPLS Core X X XX X CE PE PE CE Source Receiver P P Backup RSVP Tunnel © 2005 Cisco Systems, Inc. All rights reserved. 33
  • 34. Pseudo Wire HA PE1 PE3 PE1 P1 PE Site1 P2 P4 Site2 PE2 PE4 CE2 CE1 • Failure in the provider core mitigated with link redundancy and FRR • PE router failure–Another PE or NSF/SSO • PE-CE Attachment Circuit failure–need pair of attachment Circuits end-to-end • CE Router failure–redundant CEs © 2005 Cisco Systems, Inc. All rights reserved. 34
  • 35. Pseudowire Redundancy x PE1 PE3 CE2 PE1 P1 PE Site2 Site1 P2 P4 PE2 PE4 CE1 CE3 pe1(config)#int e 0/0.1 pe1(config-subif)#encapsulation dot1q 10 pe1(config-subif)# xconnect <PE3 router ID> <VCID> encapsulation mpls pe1(config-subif-xconn)#backup peer <PE4 router ID> <VCID> © 2005 Cisco Systems, Inc. All rights reserved. 35
  • 36. MPLS HA PE3 PE1 P1 P3 Primary Site1 x Standby P2 P4 Site2 PE2 10.1.1.0/24 PE4 10.1.2.0/24 10.1.1.0/24 10.1.2.0/24 • Link failures—solved using redundancy in network design and MPLS TE FRR and/or Fast IGP • Node failures—solved using redundant nodes and meshing of connections – Also using MPLS TE FRR node protection • Line card failures – Hardware redundancy—line card redundancy (1+1, 1:N, 1:1) • PE router (RP) failures – Dual RP systems NSF/SSO – HA software provides seamless transition… © 2005 Cisco Systems, Inc. All rights reserved. 36
  • 37. HA in MPLS L3VPN Environment RR1 RR2 10.1.1.0/24 P1 P3 10.1.2.0/24 PE1 PE1 Primary Primary Standby P2 P4 Standby 10.1.1.0/24 10.1.2.0/24 LDP iBGP • LDP and BGP use TCP as a reliable transport mechanism for its protocol messages • The TCP session between two LDP/BGP peers may go down due to HW/SW failure (RP switchover) • Use IGP, BGP, LDP NSF/GR © 2005 Cisco Systems, Inc. All rights reserved. 37
  • 38. HA in L2VPN Networks PE3 PE1 P1 PE Primary Site1 Primary Standby P2 P4 Site2 Primary PE4 • The TCP session between two LDP peers may go down due to HW/SW failure (RP switchover) • If PE3 fails, traffic will be dropped • Use PW-redundancy, LDP NSF/GR and IGP NSF/GR © 2005 Cisco Systems, Inc. All rights reserved. 38
  • 39. Agenda • IPTV Architecture overview • High Availability (HA) in the core and edge – Router Level HA with SSO – Protocol HA with NSF/GR (IGP, BGP, Multicast) – MPLS HA (LDP NSF/GR, BGP + label NSF, FRR, PW Redundancy) – Fast Convergence (Multicast FC, Fast channel change, LDP-IGP sync, LDP session protection) • Video Source Redundancy • Conclusion © 2005 Cisco Systems, Inc. All rights reserved. 39
  • 40. Fast Convergence: ☺ • Too much to cover, not possible in 45 mins • Layer 2 tuning – SONET alarms – Interface timers • CEF optimizations • IGP FC – Fast IGP timers, iSPF, fast flooding, fast hellos, etc • BGP FC – Next Hop Tracking, BGP timers • BFD (bidirectional forwarding) • Multicast Fast Convergence – Multicast timers, fast channel change • LDP – LDP (IGP sync), LDP session protection © 2005 Cisco Systems, Inc. All rights reserved. 40
  • 41. Multicast Subsecond Convergence First, what it’s NOT: A single feature or set of CLI’s So What is it? Various components seamlessly working together to provide an end-to-end fast convergence solution Objective: Achieve and meet requirements for sub- second channel change, pause, resume, etc. © 2005 Cisco Systems, Inc. All rights reserved. 41
  • 42. Multicast Subsecond Convergence: The Video Challenge - Channel Change Time & Source Redundancy Network Components: Channel Change Delay and Slow Convergence Contributors: 1 Set Top Box (STB) performance in processing channel Set Top Box change request (STB) 2 Network Signaling latency to join or resume new IP video feed 3 Delay incurred if using encryption DSLAM 4 DSLAM Signaling latency to terminate previous IP video feed Multicast Distribution 5 Network Signaling latency to rebuild tree(s) during Tree Building link/node failures in the core primary backup Video Sources 6 Time for video feed to recover during primary source failover © 2005 Cisco Systems, Inc. All rights reserved. 42
  • 43. Multicast Subsecond Convergence Areas Addressed Today: 1 Set Top Box (STB) performance in processing channel change request • Vendor dependent 2 Network Signaling latency to join or resume new IP video feed • Multicast Fast Join/Leave for Fast Channel Change 3 Delay incurred if using encryption • Vendor dependent 4 DSLAM Signaling latency to terminate previous IP video feed • Vendor dependent 5 Network Signaling latency to rebuild tree(s) in the core • IGP timers • RPF Interval timer - ip multicast rpf interval <seconds> [list <acl> | route-map route-map>]) • PIM RPF Failover timer - ip multicast rpf backoff <min> <max> [disable] • PIM Router Query interval timer - ip pim query-interval <period> [msec] • PIM Triggered Join(s) • IGMP High Availability 6 Time for video feed to recover during primary source failover • Anycast Sources (and other related methodologies) © 2005 Cisco Systems, Inc. All rights reserved. 43
  • 44. Fast Join/Leave for Faster Channel Change Problem Description: In networks where bandwidth is constrained between multicast routers and hosts (like in xDSL deployments), fast channel changes can easily lead to bandwidth oversubscription, resulting in a temporary degradation of traffic flow for all users. Solution: Reduce the leave latency during a channel change by extending the IGMPv3 protocol. Benefits: • Faster channel changing without BW oversubscription • Improved diagnostics capabilities © 2005 Cisco Systems, Inc. All rights reserved. 44
  • 45. Multicast Fast Join/Leave for Faster Channel Change How it works: “David” “John” • Relies on IGMPv3 • Router tracks both User and Channel(s) being watched igm p p • When user leaves channel no one igm else is watching, router immediately prunes the channel off the interface igmp igmp compared to IGMPv2 (up to 3 E0 seconds) and IGMPv1 (up to 180 seconds)! Configuration: interface Ethernet 00 interface Ethernet Int Channel User ip pim sparse-mode ip pim sparse-mode ip igmp version 33 ip igmp version E0 10.0.0.1, 239.1.1.1 “David” ip igmp explicit-tracking ip igmp explicit-tracking E0 10.0.0.1, 239.2.2.2 “John” First introduced in 12.0(29)S E0 10.0.0.1, 239.3.3.3 “David” © 2005 Cisco Systems, Inc. All rights reserved. 45
  • 46. Fast Convergence: LDP-IGP sync => LDP Down but IGP Still Up 1. Data stream between CE1 and CE2 traversing SP cloud 2. LDP Adjacency goes down between P4 and P3 but IGP still points to P3 as the best path 3. LDP Adjacency goes down between P4 and P3 but IGP still points to P3 as the best path RR1 RR2 P2 PE1 P1 P3 PE4 Primary P4 Primary Standby Standby Network x Network y CE1 CE2 © 2005 Cisco Systems, Inc. All rights reserved. 46
  • 47. LDP down but IGP Still Up with IGP-LDP Synch Feature 1. Data stream between CE1 and CE2 traversing SP cloud 2. LDP Adjacency goes down between P4 and P3 but IGP still points to P3 as the best path 3. P1 chooses P2 as the preferred IGP path because of the lower metric RR1 RR2 P2 PE1 P1 P3 PE4 Primary P4 Primary Standby Standby PE2 PE3 Network x Max-metric Network y advertised CE1 CE2 © 2005 Cisco Systems, Inc. All rights reserved. 47
  • 48. IGP-LDP Sync • IGP-LDP Synch features synchronizes the state of IGP and LDP between two routers • If an LDP adjacency between two routers goes down then these routers would advertise a max-metric for the link connecting them via IGP • This way upstream router can choose an alternate path if available • Following configuration needs to be enabled on the routers – P4(config)# router ospf 1 – P4(config-router)# mpls ldp sync – P4(config)# mpls ldp igp sync holddown <msecs> – If no holddown is configured then no timeout is applied i.e. IGPs will wait forever. © 2005 Cisco Systems, Inc. All rights reserved. 48
  • 49. Fast Convergence: LDP Session Protection • There are two discovery mechanisms for LDP peers 1. LDP Basic Discovery—Discovery of directly connected neighbors via link hellos 2. LDP Extended Discovery—Discovery of non-directly connected neighbors via Targeted Hellos • If the session protection is enabled, the establishment of a directly connected LDP session triggers Extended Discovery with the neighbor • With targeted hellos, session stays up even when the link goes down • No need to re-establish an LDP session with the link neighbor and relearn prefix label bindings when the link recovers p1(config)#mpls ldp session protection ? duration Period to sustain session protection after loss of link discovery P2 Extended for Access-list to specify LDP peers vrf VRF Routing/Forwarding instance information Discovery P1 P3 P4 Basic mpls ldp session Discovery protection © 2005 Cisco Systems, Inc. All rights reserved. 49
  • 50. Agenda • IPTV Architecture overview • High Availability (HA) in the core and edge – Router Level HA with SSO – Protocol HA with NSF/GR (IGP, BGP, Multicast) – MPLS HA (LDP NSF/GR, BGP + label NSF, FRR, PW Redundancy) – Fast Convergence (Multicast FC, Fast channel change, LDP-IGP sync, LDP session protection) • Video Source Redundancy • Conclusion © 2005 Cisco Systems, Inc. All rights reserved. 50
  • 51. Video Source Redundancy : Two Approaches Primary-Backup “Heartbeat” Hot-Hot Two sources, One active (src’ing content), Two sources, both are active and src’ing Second in standby mode (not src’ing content) multicast into the network Heartbeat mechanism used to communicate with No Protocol between the two sources each other Only one copy is on the network at any instant Two copies of the multicast packets will be in the network at any instant Single Multicast tree is built per the unicast routing table Two Multicast trees on almost redundant Infrastructure Uses required bandwidth Uses 2x network bandwidth Receiver’s functionality simpler: Receiver is smarter: Aware of only one src, fail-over logic handled Is aware/configured with two feeds (s1,g1), between sources. (s2,g2) / (*,g1), (*,g2) Joins both and is always receiving both feeds This approach requires the network to have fast This approach does not require fast IGP and PIM IGP and PIM convergence convergence © 2005 Cisco Systems, Inc. All rights reserved. 51
  • 52. Video Source Redundancy: Heartbeat Model Source Service Edge National Backbone Regional Backbone Residence Primary X Source 1 Heartbeat Regional Backbone Secondary Source 1 P P Primary PE Source 2 PE P P PE Heartbeat Regional Backbone PE P P PE PE Secondary PE Source 2 P P Primary Source 3 Regional Backbone Heartbeat Secondary Source 3 © 2005 Cisco Systems, Inc. All rights reserved. 52
  • 53. Video Source Redundancy: Hot-Hot Video Delivery Model Source Service Edge National Backbone Regional Backbone Source 1 P Source 1 PE P PE P P PE PE PE PE PE PE P P PE Source 2 PE PE Source 2 P PE P Source 3 Source 3 © 2005 Cisco Systems, Inc. All rights reserved. 53
  • 54. Multicast Source Redundancy Using Anycast Sources How is source redundancy achieved in the network? Anycast Sources • Enable SSM on all routers X X SF NY • Have R1 and R2 advertise same 1.1.1.1 1.1.1.1 prefix for each source segment. • R3 and R4 follow best path R1 R2 towards source based on IGP v2 join metrics. • Let’s say R3’s best path to SF is R3 R4 through R1. The source in SF now suddenly fails. I will send join • R3’s IGP will reconverge and to the nearest 1.1.1.1/32 trigger SSM joins towards R2 STB STB in NY. © 2005 Cisco Systems, Inc. All rights reserved. 54
  • 55. Conclusion • High Availability involves many more components in Metro Ethernet, DSL/PON deployments, Authentication/Application Servers – Distributed DSLAMs • Slowest component is a bottleneck for the entire system • Careful when using High Availability versus Fast Convergence. Race Conditions possible!! © 2005 Cisco Systems, Inc. All rights reserved. 55
  • 56. RST-3108 11055_05_2005_c1 © 2005 Cisco Systems, Inc. All rights reserved. © 2005 Cisco Systems, Inc. All rights reserved. 56 56