Introduction To MPLS: Technology Tutorials

Download as pdf or txt
Download as pdf or txt
You are on page 1of 50

Session 1801

Introduction to MPLS
Technology Tutorials

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

Agenda
Introduction MPLS Fundamentals MPLS Applications
Traffic Engineering Resiliency and restoration MPLS-based VPNs

Advanced Topics MPLS Support in OPNET Conclusion

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

What is MPLS?
Different things to different people One answer
Generic tunneling mechanism Evolving suite of IETF standard/near standard protocols for the Internet backbone Enabling technology for new and converged IP services Integrates packet switching with network layer routing De-couples routing from forwarding in an IP network Works with any routing paradigm Employs a simple forwarding paradigm called label swapping

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

Origins
Mid 90s Switch when you can, route when you must Bring L2 performance to L3 (IP)
Switching (L2) Simple table lookup Could be done in hardware at wire speed IP Routing (L3) Longest prefix match algorithm Was performed in software at < wire speed

Make IP networks work more like ATM without the cost and complexity

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

MPLS Timeline

Precursors started in mid 1990s


Toshiba (Cell Switching Router) Ipsilon (IP Switching) Cisco (Tag Switching) IBM (Aggregate Route-based IP Switching)

IETF MPLS working group formed in 1997


MPLS was chosen as a generic name for the technology

MPLS RFCs released in 2001


Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

MPLS Combines Routing and Switching

IP Routing

ATM Switching

Multiprotocol Label Switching

IP routing (pure Layer 3 technology)


Provides rich functionality: wide range of protocols, interface types, and speeds

ATM switching (pure Layer 2 technology)


Does simple forwarding of Layer 2 protocol packets based on circuit numbers

One view is that MPLS combines the best of both


Rich functionality and flexibility of Layer 3 routing Speed and simplicity of Layer 2 switching

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

Motivation for MPLS Today


Original performance motivations no longer relevant
LPM can be done at wire speed

Other factors have taken over


Growth and evolution of the Internet Growing number of users Increasing need for bandwidth Diverse service types and QoS requirements Use of overlapping address space (RFC 1918) Managing bandwidth vs. buying bandwidth Limitations of existing core technologies Movement to a single unified network

Need for scalability in the Internet backbone

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

Agenda
Introduction MPLS Fundamentals MPLS Applications
Traffic Engineering Resiliency and restoration MPLS-based VPNs

Advanced Topics MPLS Support in OPNET Conclusion

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

MPLS Fundamentals
How it works The Label Switched Path (LSP) Label Switching Router (LSR) functions Traffic assignment Inside the MPLS label

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

Label Switching Router (LSR)


Sets up Label Switched Paths (LSPs) Forwards traffic along LSPs using label swapping Can be a router or switch Runs one or more IP routing protocols
to learn network topology to distribute MPLS topology state information to other LSRs to forward native IP packets

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

10

1801 Introduction to MPLS

Label Switched Path (LSP)


A unidirectional tunnel through the MPLS domain
For a round trip, two LSPs are required

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

11

1801 Introduction to MPLS

Forwarding Equivalence Class (FEC)


Definition: A group of IP packets that are forwarded in the same way Packets are classified into FECs
Only once At the ingress to the MPLS domain

A FEC identifies a set of IP packets to map to an LSP Packets in the same FEC
Receive the same label from the ingress LSR Are mapped to the same LSP and forwarded over the same path (or sets of paths in the case of multi-path routing)

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

12

1801 Introduction to MPLS

FEC (cont.)
FECs are not necessarily new
In conventional IP, a FEC is formed at each router based on Layer 3 lookup Packets with the same longest matching address prefix (based on destination address) are treated in the same way

FECs are currently derived from IP routing protocols


Based on destination IP prefix (IP header) Mappings can be policy-based (e.g., ToS bits)

MPLS offers additional flexibility and granularity for classification of FECs, such as
Same egress router or switch Same longest matching destination address IP prefix Same longest matching destination IP Prefix AND same Type of Service bits Same application flow

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

13

1801 Introduction to MPLS

MPLS How It Works


LSRs use (extended) link state IGPs to learn network topology Path setup: For each LSP configured on an ingress LSR:
Ingress LSR looks up or calculates the path Ingress LSR signals the LSP Transit and egress LSRs set up labels for the LSP and confirm to ingress LSR

Forwarding: For each packet that arrives on an ingress LSR:


Ingress LSRs assigns traffic to LSPs based on FEC Interior LSRs forward traffic using label switching Egress LSR forwards traffic based on IP or VPN rules

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

14

1801 Introduction to MPLS

MPLS Domain Boundaries

IP packet enters the MPLS domain Ingress LSR (LSR1) assigns a label and forwards the packet to the next hop in the label switched path (LSP) Intermediate LSR (LSR2, LSR3) does a simple lookup, swaps the label, and forwards the packet Egress LSR (LSR4) or Penultimate hop (LS3) removes the label and forwards the packet using based on conventional IP or VPN rules
Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

15

1801 Introduction to MPLS

Path Setup Example


LSR1 transmits a Label Request message to LSR4
Each downstream router modifies the route list

LSR4 transmits a Label Mapping message to LSR1


LSR4 assigns an inbound label and transmits upstream

Intermediate LSRs (LSR3 and LSR2)


Store outbound label provided by downstream LSR Assign an inbound label and transmit upstream

LSR1 binds the label to the FEC

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

16

1801 Introduction to MPLS

Packet Forwarding Example


Ingress: LSR1 Egress: LSR 4

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

17

1801 Introduction to MPLS

MPLS Label
Short fixed length identifier used to designate a FEC Has local significance only Changes from hop to hop For IP, the label is contained in a shim header
For ATM the label is VPI/VCI For Frame Relay the label is DLCI

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

18

1801 Introduction to MPLS

MPLS Packet Format and Shim Header


MPLS is often described as introducing a shim header between the original layer 2 and layer 3 headers
This is the reason MPLS is sometimes described as Layer 2.5

The 32-bit MPLS shim header is added to the IP header


Maps network layer routing to data link layer switched paths

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

19

1801 Introduction to MPLS

Where Does MPLS Fit in the OSI Model?


MPLS works with and supports Layer 3 technologies, but does not have routing and addressing
MPLS is not Layer 3 MPLS is not Layer 2

MPLS is Layer 2.5 Shim Layer


It helps Layer 2 and Layer 3 fit better

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

20

1801 Introduction to MPLS

Label Stacking
Labels can be ordered hierarchically in a stack Label stacks permit nesting of LSPs
Similar to ATM VPs for aggregating multiple VCs, but MPLS supports arbitrary levels of hierarchy

Can be used to reduce the number of LSPs through the core Only top label is swapped
Packets are forwarded based on the value of the label at the top of the stack Last-in, first-out stack

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

21

1801 Introduction to MPLS

Label Stacking Example


Useful for IP/MPLS VPNs and TE (illustrated later) Also used to support resiliency (FRR bypass tunnels)

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

22

1801 Introduction to MPLS

MPLS Routing
Topology Determination Path Determination
IGP CSPF Explicit Routing

IP Routing Interactions Load Balancing

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

23

1801 Introduction to MPLS

Topology Determination
Definition: An MPLS domain is a set of physically connected LSRs (includes LSRs acting as LERs) Routers within an MPLS domain use routing protocols to discover the network topology MPLS IGPs: OSPF-TE and ISIS-TE MPLS EGP: BGP4

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

24

1801 Introduction to MPLS

LSP Path Determination


Path determination options depending on label distribution protocol LSP paths can be determined using
LDP: Routers calculate dynamically using IGP Selects IGP shortest path RSVP-TE: Routers calculate dynamically using CSPF Selects shortest path that meets constraints RSVP-TE: Network operator specifies using Explicit Routes (ERs) Uses configured ERs Multiple Explicit Routes can be configured per LSP Primary (no more than one) Secondary (zero or more)

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

25

1801 Introduction to MPLS

LSP Attributes
Path Definition
Defined at ingress LSR Remote destination (usually loopback address)

Path Selection and Management


Administratively configured explicit routes Explicit routes may be mandatory Fallback to CSPF CSPF constraints, including: Required bandwidth Maximum hop count Resource classes: eligibility to use a link Must be consistent with resource classes configured on interface

Priority (Setup and Holding)


Used for preemption (policy-based bumping) in dynamic routing

Resilience Mode (Recovery policy)


Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

26

1801 Introduction to MPLS

Constraint-Based Shortest Path First (CSPF)


Automated constraint-based TE is its intent
Associate flow requirements with a FEC Track new link state parameters TE Extensions to OSPF and IS-IS Calculate the shortest path across the MPLS domain that Meets the flow requirement based on current network state Meets a set of constraints specified in LSP attributes

Path cost based on Dijkstras shortest path first (SPF) algorithm


Build a network graph Graph edge (link) cost: inherit or override IGP link cost Apply constraints: prune a link if Insufficient resources to accommodate the LSP Link cannot satisfy LSP local constraints (e.g. resource classes) Compute shortest (least-cost) path using the pruned graph Path must also satisfy LSP constraints (e.g. maximum hops)
27

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

Explicit Routing
Explicitly routed LSPs
Sometimes referred to as Traffic Engineered Tunnels

Administratively pinning routes of LSPs


Done manually or automatically (e.g., using a TE computation)

Can mix and match with dynamically routed LSPs


Local (selected LSPs, partial mesh) a.k.a. Tactical Global (full mesh among LERs)

Permits centralized, global decision making for traffic engineering


Explicit Routes are the output (decision variables) of TE

Indirectly enables QoS- and service-level-focused mechanisms


Assuring that certain traffic or service types traverse certain network resources (devices, links)

Possibly computed using external TE solution


28

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

Explicitly Routed LSPs


Explicit routes can be strictly or loosely defined
Strict: All hops are specified from ingress to egress, that is, each next hop is directly connected (fully pinned) Loose: The path between ingress and egress is partially specified (partially pinned). When the next hop is not directly connected, use IGP or CSPF to reach it.

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

29

1801 Introduction to MPLS

Agenda
Introduction MPLS Fundamentals MPLS Applications
Traffic Engineering Resiliency and restoration MPLS-based VPNs

Advanced Topics MPLS Support in OPNET Conclusion

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

30

1801 Introduction to MPLS

Traffic Engineering
Top-level view
Capacity Planning: placing bandwidth to support traffic Traffic Engineering: placing traffic where there is bandwidth

MPLS ability to arbitrarily segregate flows at whatever level of granularity is desired and to route those flows independently of one another (regardless of source/destination addresses) forms the basis for traffic engineering

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

31

1801 Introduction to MPLS

MPLS Traffic Engineering


MPLS traffic engineering defined
Definition Controlling traffic in a predictable manner to maintain service levels Goal Optimize network resource utilization and traffic performance

Three types
Inline Online Offline TE performed on a device using local information TE done using global information by a central server connected to the network TE done by a server external to the network using global information

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

32

1801 Introduction to MPLS

Why TE?
Bandwidth availability
Infrastructure limitations, lead times

Pipe size granularity issues Class-of-service routing Knobs to tweak under failure scenarios Hedge against traffic issues
Uncertainty, growth, fluctuations

Economics
Especially today

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

33

1801 Introduction to MPLS

Traditional IP TE Motivation
Problem: Hyper-aggregation of flows

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

34

1801 Introduction to MPLS

Traditional IP TE Cycle
Solution approach: Trial and Error

Classically Unstable
Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

Still flawed, but less so with predictive tools


35

1801 Introduction to MPLS

The Problem with Traditional IP TE


Brute force solution

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

36

1801 Introduction to MPLS

TE with MPLS
MPLS Tactical LSP Solution

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

37

1801 Introduction to MPLS

MPLS Traffic Engineering Online/Offline


MPLS provides the building blocks to perform IP traffic engineering better, but it does not provide the full TE solution TE presents an opportunity to solve some global optimization problems focused on balancing loads and improving service levels This requires new TE software, methodology, and processes

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

38

1801 Introduction to MPLS

MPLS Online/Offline TE Process

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

39

1801 Introduction to MPLS

MPLS TE Automated Model-Building


Automatically constructing a detailed, operationally correct model of the existing network
Topology (nodes and links) Detailed device and protocol configuration Existing LSPs, their configuration, routes Link and LSP usage information IF-MIB (Cisco), IF-MIB extension (Juniper) (Optionally) traffic Usual imperfect sources 3rd party systems TMS (Cisco) Traffic inference

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

40

1801 Introduction to MPLS

MPLS TE Explicit Route Generation


Automated design and analysis of traffic engineering solutions against operational goals
Design CSPF versus explicit routing Explicit route computations (primary, secondary, restoration, etc.) Analysis Performance analysis (e.g., design utilization metrics, device and link usage/subscription metrics, delay metrics, etc.) Failure analysis Traffic growth analysis Topology analysis

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

41

1801 Introduction to MPLS

Global LSP Optimization vs. Greedy LSP Routing


Greedy: Ingress router uses the constrained shortest path at LSP setup time The setup order can greatly affect the overall solution quality Global optimization: use a holistic view to generate a globally optimal solution Example: Largest LSP (size 8) takes its shortest path, other LSPs are blocked

Routed second 6 8
6 - blocked

12 8

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

42

1801 Introduction to MPLS

MPLS TE Validation
Supports network operations in understanding and using expert judgment about the final changes to be implemented Must be supported on two levels:
Summary reports on MPLS configuration adds, deletes, or changes and their impact on design criteria and operational tolerances Ability to directly review and diff configurations for the affected devices

Validation concerns include:


Correctness Value of changes Ensuring that decisions were based on accurate and current data

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

43

1801 Introduction to MPLS

MPLS TE Deployment
Ability to parse the validated configuration results generated by the system into a form most useful for implementation Issues here are:
Deployment model (matter of operations preference) Granularity, order, chunking Deployment means Direct through device configlets, SNMP, NMS/OSS interfaces

Requires Change Management functions consistent with deployment model and means
Ability to introduce, check point, archive, and back out configuration changes

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

44

1801 Introduction to MPLS

MPLS TE Automating the Process


The answer to What is the appropriate time scale for this cycle? drives automation Closer to being a reality in the technology than one that will be accepted organizationally
IP/Optical and other NGN initiatives may contribute to accelerating the technology and increasing its acceptance

Expect a gradual transition through


Human operated At each process step Human supervised For validation and to supervise deployment Exception managed Operated like IGPs are today

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

45

1801 Introduction to MPLS

MPLS TE Results on an Example Network


Basic MPLS TE load balancing via primary ERs improves network performance Survivable TE assures network failure resilience
Maximum Link Utilization 100% 80% 60% 40% 20% 0% CSPF 96% 41% TE 39% Failure Normal

Note: Results are network-specific.


Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

46

1801 Introduction to MPLS

Summary: TE Options
Inline (CSPF)
+ Still better than IGP routing + Least overall complexity + No need for external TE system - Non-optimal use of bandwidth - Still need process or mechanism to size LSPs - Vendor interoperability issues?

Online/Offline TE (ERs)
+ Most efficient use of bandwidth + Better protection (SRGs) - Can be operationally complex

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

47

1801 Introduction to MPLS

MPLS TE Deployment Considerations


Governed by underlying topology, traffic mix and applications
MPLS topology different deployment models for LSP topology (flat/hierarchical) Flow segregation different strategies for flow segregation onto LSPs (FECs) Application-specific deployment to support specific applications or services (QoS/ToS per-hop behaviors)

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

48

1801 Introduction to MPLS

MPLS Topology For Traffic Engineering


For TE purposes, MPLS is deployed in the core routers (or a TE layer internal to the core routers) Deployment scenarios include
Tactical deployment to fix a particular problem Alleviate congestion Improve service level(s) Fully traffic-engineered flows Motivated by measurement it enables and control Full-mesh or hierarchical

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

49

1801 Introduction to MPLS

MPLS Topology For Tactical TE


To alleviate congestion, an LSP is created to move one of the flows on the congested link to an alternate (non-IGP) route

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

50

1801 Introduction to MPLS

MPLS Topology A Full TE Mesh


Enables measurement octet/packet counts on each LSP Enables control routing decisions per LSP if needed

Flat Deployment

Hierarchical Deployment

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

51

1801 Introduction to MPLS

Intermission

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

52

1801 Introduction to MPLS

Agenda
Introduction MPLS Fundamentals MPLS Applications
Traffic Engineering Resiliency and restoration MPLS-based VPNs

Advanced Topics MPLS Support in OPNET Conclusion

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

53

1801 Introduction to MPLS

MPLS Resiliency and Restoration


An LSP becomes unusable if any network resource along its route fails LSP restoration mechanisms can be setup at different time scales
Mechanisms generally have a tradeoff between the time required to restore service after a failure, resources used, and complexity of configuration Slower mechanisms tend to provide better long-term solutions in terms of network resources Faster mechanisms protect in-flight data but at the cost of sub-optimal use of network resources Some carriers seeking near SONET (50 milliseconds) restoration times Multiple mechanisms make sense

A networks resiliency is the degree to which the network can successfully survive failures

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

54

1801 Introduction to MPLS

Resiliency and Restoration


Can occur at one or several layers
Optical layer SONET layer MPLS layer IP layer Routing protocol convergence

Configuring restoration mechanisms at all layers can be expensive Need to balance cost and complexity of planning for resiliency with cost and risk of a failure.

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

55

1801 Introduction to MPLS

Types of Failures
Link Failures Node Failures Shared Risk Group (SRG) Failures
SRGs are collections of network resources that share the same risk of failure. Examples: Circuits that traverse that same physical fiber span (fiber cut) Devices in the same building (natural disaster) Devices sharing the same power supply (power failure)

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

56

1801 Introduction to MPLS

MPLS Restoration Two Common Means


Path protection
Head end can reroute what it signaled Longer-term, more optimized, repair made at the source Motivation is quality of repair at a cost of speed ~ O(seconds) (Alternative strategy: have an alternate LSP up and running whose usage under normal conditions is precluded using metrics)

Local protection
Temporary, likely sub-optimal, repair made locally in the neighborhood of the point of failure to keep critical flows up Motivation is speed ~ O(milliseconds) Attempt to keep data in flight until more permanent repair can be made Example: Fast Reroute

Path and local protection are complementary


One is a short term fix, the other a long(er) term fix
57

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

Path Protection

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

58

1801 Introduction to MPLS

Link Protection (Local)

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

59

1801 Introduction to MPLS

Node Protection (Local)

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

60

1801 Introduction to MPLS

SRG Protection (Local)

Protected SRG

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

61

1801 Introduction to MPLS

MPLS Protection Approaches


Path protection
Failover to CSPF route Default Requires head-end router to detect failure, recompute shortest path on the remaining network, and set up new path (may be several seconds) Failover to precomputed secondary route Requires head-end router to detect failure and set up new path The secondary route should be failure disjoint from the primary Secondary route only uses resources when the primary fails Failover to backup (standby) LSP For each primary LSP, one or more backup LSPs are designated Backup LSPs are set up before failures occur and can consume resources under nonfailure conditions Can be set up with zero bandwidth TE metric used to prevent use of the backup LSP under non-failure conditions Head-end router switches from primary to backup when it detects the failure

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

62

1801 Introduction to MPLS

MPLS Protection Approaches


Local protection
Each LSR in the path has a precomputed alternate next-hop LSP to replace the physical next hop if the primary becomes unavailable (Cisco Fast Reroute) Requires stackable LSPs (LSPs riding other LSPs) Does not require head-end signaling (45-50 milliseconds typical) Does not use additional resources until the failure occurs Temporary solution until head-end router can restore the LSP

Physical layer protection


Relying on the SONET redundancy features to handle link failures before they are detected by IP/MPLS (< 50 milliseconds)

Hybrid strategies
Example protection strategy: Platinum/Real-time traffic (VoIP/Video): FRR Gold/Premium: secondary explicit routes Bronze/Best effort: no protection
Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

63

1801 Introduction to MPLS

Agenda
Introduction MPLS Fundamentals MPLS Applications
Traffic Engineering Resiliency and restoration MPLS-based VPNs

Advanced Topics MPLS Support in OPNET Conclusion

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

64

1801 Introduction to MPLS

MPLS-Enabled IP VPNs
Head-to-head with MPLS TE in importance MPLS VPNs (Virtual Private Networks) are inherently based on MPLS ability to segregate flows in this case on a per VPN (i.e. per customer) basis from provider edge (PE) to provider edge (PE) Key motivators (analogous to FR/ATM) for MPLS VPNs
Revenue Address space reuse and overall ease of management, security, etc. Ability to address customer service levels (via routing or in combination with QoS mechanisms) and monitor customer traffic Granularity of decisions available under failure conditions

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

65

1801 Introduction to MPLS

MPLS-Based VPNs
Motivation for MPLS VPNs MPLS-Based Layer 2 VPNs MPLS-Based Layer 3 VPNs Tradeoffs MPLS-Based Layer 2 versus Layer 3 VPNs

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

66

1801 Introduction to MPLS

Motivation for MPLS L2 VPNs


Have a single network technology for all types of services PE-to-PE regardless of the customer-facing technology (decouple PE technology from CE technology)
One operations center, reduced staff, one OSS/BSS infrastructure

A single MPLS infrastructure enables traditional (FR, ATM) and new (Ethernet) VPN services over a single Packet-over-SONET (POS) infrastructure
Network consolidation for SPs offering private data and IP services New revenue opportunity for IP services only providers

Simplify provisioning
Signaling and label stacking Touch only edge devices

Scalability
Core switches aggregate MPLS tunnels (label stacked) and thus manages fewer connections
Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

67

1801 Introduction to MPLS

MPLS-Based Layer 2 VPNs


Martini MPLS Layer 2 VPNs
Encapsulations for Frame Relay, Ethernet port /802.1q VLAN, ATM AAL5, ATM Cells, and PPP/HDLC Provider pre-provisions outer (service-related) LSPs all services look like a virtual circuit to the MPLS network Each service is provisioned over MPLS using LDP signaling by associating each endpoint with common VC identifier (VCID) e.g., for FR, the port/DLCI at each end is associated with the same VCID Network automatically determines VC Label to push onto the layer 2 frame LDP sessions advertise VC Labels for VCIDs Network also determines Tunnel Label to stack on top based on usual routing

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

68

1801 Introduction to MPLS

Martini MPLS Layer 2 VPNs

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

69

1801 Introduction to MPLS

Example L2 VPN Frame Relay


FR from customer premises (e.g., FRAD) to edge LSR Edge LSR
Translates FR DLCIs Maintains VC Label to in/out port and DLCI mappings

MPLS defines the label distribution and encapsulation FR PDU (including header, FECN and BECN bits, ) transported in their entirety edge to edge FR DE bit mapped to MPLS EXP values

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

70

1801 Introduction to MPLS

Example L2 VPN ATM


ATM from customer premises (e.g., ATM Switch) to edge LSR Edge LSR
Translates ATM VPI/VCIs Maintains VC label to in/out port and VPI/VCI mappings

MPLS defines the label distribution and encapsulation AAL5 and ATM cell transport modes are supported
AAL5 mode reassembles ATM PDUs from a VC into a packet Cell mode transports each ATM cell as a packet

CLP bit to EXP field mapping supported

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

71

1801 Introduction to MPLS

Example L2 VPN Ethernet


Ethernet/FastEthernet/GigabitEthernet from customer premises (e.g., Ethernet Switch) to edge LSR Edge LSR
Translates MAC addresses Maintains MAC label to in/out port and optionally VLAN mappings

MPLS defines the label distribution and encapsulation Ethernet frame is transported VLAN tags are transported Priority to EXP field mapping

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

72

1801 Introduction to MPLS

Kompella MPLS Layer 2 VPNs


Similarities with Martini VPNs
Similar approach to label stacking for scalability Similar applications (ATM, FR, Metro Ethernet)

Differences
VPN membership information distributed automatically via BGP VPN sites can be added with little provisioning BGP permits Service Provider to inter-work unlike media (e.g., ATM and FR) in a scalable fashion over MPLS Extended service offerings

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

73

1801 Introduction to MPLS

VPLS (Virtual Private LAN Services)


Martini VPNs only provide point-to-point connectivity VPLS builds upon Martini to provide multipoint connectivity Alternative to L3 MPLS VPNs Ethernet based (Virtual LAN)
Per-customer broadcast domain Full mesh of Martini tunnels between PE devices PE devices learn MAC forwarding information just like regular Ethernet switch Frames with unknown MAC addresses are broadcast

Full mesh and broadcast nature of Ethernet creates scalability issues


Hierarchical-VPLS (H-VPLS) addresses these limitations 2 tier architecture

Draft-ietf-l2vpn-vpls-ldp-01.txt

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

74

1801 Introduction to MPLS

MPLS-Based Layer 3 VPNs


Mature technology based on BGP/MPLS VPNs
RFC2547

Services enabled
IP VPNs to enterprise customers Inter-provider VPNs hook two VPNs together across providers Carrier-of-carrier services IP transport to retail ISPs BGP/MPLS VPN across carrier core only IP transport to SP itself providing L2/L3 services BGP/MPLS VPN across the network of SP and carrier

Mature technology
Large-scale deployments Hardware optimized for scalability in excess of 1000 VPNs per PE Mature provisioning/management software

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

75

1801 Introduction to MPLS

MPLS Topology for MPLS BGP VPNs


VPNs with MPLS and BGP
Internet Draft draft-rosen-rfc2547bis-03.txt (Feb 2001)

Three device roles are defined


CE (customer edge) Router PE (provider edge) LSR P (provider core) LSR

PE device:multiple virtual routing/forwarding (VRF) tables


One forwarding table per set of directly attached sites with common VPN membership Customer routes are extended with unique label (Route Distinguisher) Permits private addressing Multiprotocol BGP (MBGP) extensions advertise VPN reachability PE LSRs participate in a full mesh of MBGP that distributes VPN labels

LDP typically used to distribute path labels from PE-to-PE routers


Uses MPLS hop-by-hop routing along IGP path P routers do not need to be aware of VPN routes
Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

76

1801 Introduction to MPLS

RFC 2547: Forwarding Plane

IP

IP

2547 L2

MPLS L1 MPLS

CE PE 1) Receive IP and send IP datagram to PE via ATM, FR, Ethernet, etc. 2) Add RFC 2547 Header Label for VPN ID. Add MPLS tunnel; label and send to MPLS network.
77

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

RFC 2547: Forwarding Plane


CE CE

IP

2547 L2

MPLS L1 MPLS IP PE

3) Pop MPLS tunnel label. 4) Pop VPN label and send to CE.

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

78

1801 Introduction to MPLS

RFC 2547: Control Plane


VRF Blue
2) PE 1 build VRF for

CE 1 PE 1 P

VRF BLUE 192.168.10.0/24.

1) CE1 PE 1 Exchange routers with IGP (Rip, OSPF, IS-IS) 192.168.10.0/24.


Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

79

1801 Introduction to MPLS

RFC 2547: Control Plane


A PE 1 CE 1 P PE 2
3) PE1 PE 2 Exchange routes for Blue VPN with BGP 192.168.10.0/24. Do not share with P routers. Use LDP tunnel or RSVP.
80

VRF Blue

CE 2

4) PE 2 build VRF Blue VPN for 192.168.10.0/24.

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

RFC-2547: Overlapping Private Addresses


192.168.10.0/24 VRF Blue Company A 192.168.10.0/24

CE CE

VRF Red Company B 192.168.10.0/24 192.168.10.0/24


81

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

RFC-2547: Overlapping Private Addresses


192.168.10.0/24 BGP RD 1 (blue) 192.168.10.0/24 CE CE RD 2 (red) 192.168.10.0/24 192.168.10.0/24

Company B 192.168.10.0/24 192.168.10.0/24


82

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

MPLS L2 VPNs Versus L3 VPNs


L2 VPNs (Martini/Kompella/VPLS) Positives:
Traditional L2 VPN from customers perspective Provider not routing customer traffic Single network architecture and infrastructure for both Internet and VPN traffic Decouples core and edge technologies Auto-provisioning via LDP setup

L3 VPNs (RFC2547) Positives:


Value-added service for customers that want to outsource Mature technology Lots of (somewhat esoteric) service opportunities QoS/CoS, carrier of carriers, inter-SP VPNs

Negatives:
Point-to-point focus (Martini/Kompella) Scalability (VPLS) Not as flexible in terms of service opportunities

Negatives:
Not transparent - migration requires effort Customer must peer with provider CE device must be a router Some customers strongly object to this invasion of privacy
83

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

Agenda
Introduction MPLS Fundamentals MPLS Applications
Traffic Engineering Resiliency and restoration MPLS-based VPNs

Advanced Topics MPLS Support in OPNET Conclusion

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

84

1801 Introduction to MPLS

Advanced Topics
Implementing QoS in MPLS IP Routing Interactions IGP Interactions Load Balancing Status of MPLS Whos working on MPLS

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

85

1801 Introduction to MPLS

Implementing QoS in MPLS


Multiple service levels (e.g., Bronze, Gold, Platinum) Service Level assignment based on VPN (ingress port) or ToS (IP header)

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

86

1801 Introduction to MPLS

Implementing QoS in MPLS LSP-based


Strategy 1: Apply QoS to LSP
Multiple LSPs between each ingress/egress LER (full mesh per service level!) Destination IP address & ToS, or VPN, used in FEC L-LSP LSPs differentiated by Setup/Hold Priorities (for dynamic/CSPF routing) Primary Explicit Routes (favoring some LSPs in global optimization) Protection mechanisms (Fast Reroute, Secondary Explicit Routes) Resource classes (to reserve shortest paths for best service)

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

87

1801 Introduction to MPLS

Implementing QoS in MPLS IPQoS-based


Strategy 2: Piggyback underlying IP QoS
Single LSP between each ingress/egress LER Destination address (only) used in FEC E-LSP Use ToS to assign EXP bits in MPLS Shim header Configure transit LSRs to provide favorable queuing based on EXP bits Must provide protection mechanisms (Fast Reroute, Secondary ERs) and adequate bandwidth (primary and protection) to all LSPs

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

88

1801 Introduction to MPLS

Implementing QoS in MPLS DiffServ TE


Strategy 3: DiffServ TE OPNETWORK 1825 Advanced Topics in MPLS: QoS, DiffServ TE, and GMPLS

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

89

1801 Introduction to MPLS

IP Routing Interactions
In an operational network, routing can be configured in a number of ways so that flows are routed using LSPs
BGP ingress/egress mode Flows entering the network at an AS boundary can have their BGP next hop set to point to an LSP Mechanism used for L3 MPLS VPNs IGP Shortcut LSP Examples are Ciscos Autoroute and Juniper IGP Shortcuts Visible at head-end LER only After IGP routing has computed the shortest path tree, a post processing step is used to replace IGP next hops with shortcut LSP paths Forwarding Adjacency LSPs Directly used in the IGP shortest path computation as layer-3 adjacencies More predictable and intuitive than shortcuts Results in N2 adjacencies in an LSP mesh
Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

90

1801 Introduction to MPLS

IGP Interactions
IGPs often support equal-weight split path routing at each hop along the IGP path to a destination
The number of splits per hop is small typically four, but it is configurable The number of splits compounds geometrically hop-to-hop (4x4x4, ) This creates de-facto load balancing under the best of circumstances Can also create congestion where the equal-weight paths (IGP link weights are configured) do not reflect the link capacities along the paths

MPLS deployment disables IGP split pathing MPLS can be configured similarly to provide split path routing along parallel LSPs
Splitting is proportional to LSP bandwidth Splitting occurs only once at the ingress of the parallel LSPs

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

91

1801 Introduction to MPLS

Load Balancing
There are two categories of load balancing in MPLS
Path selection When multiple equal cost paths to egress are available, CSPF can use tie breaking rules to select the one to use: Random randomly select a path to use Least-fill prefer the path with the largest minimum available bandwidth ratio Most-fill prefer the path with the smallest minimum available bandwidth ratio where available bandwidth ratio = (avail bw on link)/(max reservable bw on link) Balancing traffic over multiple LSPs Per-prefix (IP addr/netmask) keeps individual flows on one route Per-packet can split individual flows over multiple LSPs in proportion to the bandwidth of the LSPs

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

92

1801 Introduction to MPLS

Status of MPLS
Lots of excitement Hundreds of deployments worldwide
Cisco reported >200 deployments in 2003 Almost all providers offing some form of MPLS VPN service Most are doing TE within their core

Standardization work continues


RFCs, internet drafts

Interoperability labs
University of New Hampshire's InterOperability Lab Isocore Internetworking Lab

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

93

1801 Introduction to MPLS

Whos Working on MPLS?

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

94

1801 Introduction to MPLS

Agenda
Introduction MPLS Fundamentals MPLS Applications
Traffic Engineering Resiliency and restoration MPLS-based VPNs

Advanced Topics MPLS Support in OPNET Conclusion

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

95

1801 Introduction to MPLS

OPNET Support for MPLS?


MPLS data collection
Routers, LSPs, configuration LSP utilization Cisco, Juniper, Foundry

MPLS modeling, simulation & optimization


CSPF (OSPF-TE, ISIS-TE), ERs LDP, RSVP QoS, Diffserv-TE Failure analysis Traffic engineering optimization Resiliency design

MPLS VPNs
L2 (Martini, Kompella, VPLS) & L3 (RFC 2547) Graphical provisioning wizard Views to study logical VPN topology

Support for MPLS-related R&D


Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

96

1801 Introduction to MPLS

Agenda
Introduction MPLS Fundamentals MPLS Applications
Traffic Engineering Resiliency and restoration MPLS-based VPNs

Advanced Topics MPLS Support in OPNET Conclusion

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

97

1801 Introduction to MPLS

References
Books
MPLS Technologies and Applications (Bruce Davie and Yakov Rekhter, Morgan Kaufmann, 2000) Advanced MPLS Design and Implementation (Vivek Alwayn, ciscopress.com, 2002) MPLS and VPN Architectures (Ivan Pepelnjak and Jim Guichard, ciscopress.com, 2001)

Many vendors have literature posted on their websites RFC and Internet draft documents
http://www.ietf.org/html.charters/mpls-charter.html

MPLS Forum
http://www.mplsforum.org

MPLS Resource Center


http://www.mplsrc.com
98

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

1801 Introduction to MPLS

Other MPLS-Related Sessions


Network Tutorials Track
1818 Introduction to VPNs 1825 Advanced Topics in MPLS: QoS, DiffServ TE, and GMPLS

Network Analysis, Planning and Troubleshooting


1331 Planning and Analyzing VPN Architectures 1310 Planning, Analyzing, and Optimizing MPLS TE and FRR Deployments 1354 Planning, Analyzing, and Optimizing DiffServ TE and MPLS QoS

Discrete Event Simulation for R&D


1511 Understanding MPLS Model Internals

Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

99

1801 Introduction to MPLS

Take-Away Points
Main Concepts
Separates control and data plane Supports multiple routing paradigms Simple forwarding paradigm (label swapping)

Enables advanced IP Services


Triple Play (Voice, Video, and Data) with QoS Traffic Engineering Resiliency and restoration VPNs

Compatible with existing technologies


ATM, Frame Relay, Ethernet

Broadly supported in OPNET products


Import: VNE Server and MVI Simulation and design: SP Guru
Copyright 2005 OPNET Technologies, Inc. Confidential, not for distribution to third parties.

100

You might also like