A presentation given by RAD’s CTO, Dr. Yaakov Stein, at the 2012 MPLS and Ethernet World Congress. The presentation compares the two technologies in ten critical categories and grades them on suitability, coverage and maturity
Strategize a Smooth Tenant-to-tenant Migration and Copilot Takeoff
Ethernet vs-mpls-tp-in-the-access-presentation
1. Ethernet vs. MPLS-TP
in Access Networks
Yaakov (J) Stein
CTO
RAD Data Communications
Ethernet vs. MPLS-TP in the Access Slide 1
2. Agenda
Is MPLS-TP ready to replace Ethernet as the packet technology
that dominates access networks?
• Brief review of access networks
• Characteristics of Ethernet and MPLS-TP
• Technical comparison: Ethernet vs. MPLS-TP
Ethernet vs. MPLS-TP in the Access Slide 2
4. Access Network Considerations
Residential
Customers
(IP)
Access Network: Core:
Q-in-Q Ethernet? MPLS or IP
Business Customers interface
(IP or Ethernet) MPLS-TP ? (and theoretically PBB)
Cell Sites
(IP and/or Ethernet
and/or TDM Other Data
Other Internet Centers
and Timing) customer
customer
sites
sites
Ethernet vs. MPLS-TP in the Access Slide 4
5. Why Ethernet and MPLS-TP ?
First Mile
Customer Core
Network: Access Network Network:
Ethernet MPLS
Last Mile
• Ethernet started in customer networks (LAN) and over the
years has moved into the access network (MEF)
• MPLS started in the core network and is now trying to
conquer the access network
Ethernet vs. MPLS-TP in the Access Slide 5
6. Access Network Segmentation
First/Last Mile Middle Mile
NTU/CLE Access
Node
+ Aggregation
Backhaul
Access Network
A recent trend is to segment the access network into:
• First/Last Mile
– Provides connectivity from customer site to first access node
– Leverages physical layer technologies such as DSL, active/passive fiber,
microwave, HSDPA+, LTE, …
• Middle Mile
– Collects and aggregates traffic from multiple access nodes
– Provides backhaul towards core
Ethernet vs. MPLS-TP in the Access Slide 6
7. Access / Core Differences (1)
Differences between core and access networks may translate to
protocol requirements differences:
Core Access Impact
No. of Few – routers, Many – CPEs, NTUs, • Strong pressure on
Network LSRs, switches DSLAMs, access NE price levels
Elements aggregators • Access needs to be
as “touchless” as
possible
Data Rates Higher Lower • Core may guarantee
QoS by resource
over-provisioning
• Access needs QoS
mechanisms
Ethernet vs. MPLS-TP in the Access Slide 7
8. Access / Core Differences (2)
Core Access Impact
Topology Richly connected Simple – typically • Fault in access network
trees or rings affects fewer people, but
fewer bypass options
• Core can get away with
fast rerouting
• Access network requires
OAM and planned APS
Security “Walled garden”: Easily accessible • Customer networks also
• Strong security considered walled gardens
to and from • Impractical to protect the
the outside entire access network
world
• Loose security
on the inside
Ethernet vs. MPLS-TP in the Access Slide 8
10. Ethernet / MPLS-TP Differences (1)
While both Ethernet and MPLS are commonly used to carry IP,
there are some fundamental protocol differences:
Ethernet MPLS-TP
OSI Layer L0-L2 Requires a server layer to
(but may run over MPLS) transport it (which may be
Ethernet)
Packet Frames are inherently No PID
Identification self-describing
Scope Destination address: Only locally-meaningful labels
Global, not aggregated
Source Unique source address No source identifier
Identifier
Clients IP and other IP and other
Transport Over SDH/SONET, OTN Over SDH/SONET, OTN
Ethernet vs. MPLS-TP in the Access Slide 10
11. Ethernet / MPLS-TP Differences (2)
Ethernet MPLS-TP
OAM and • Fault Management • Fault Management
Protection • Performance Monitoring • Performance Monitoring
• APS • APS
IP Protocols • No routing protocol Leverages the entire IP suite
defined of protocols
• A number of L2CPs
Loop No Can tolerate transient loops
Tolerance (has TTL field)
Priority • 3-bit (e.g., DiffServ) 3-bit (DiffServ)
Marking • “Drop Eligibility”
Additional • Bandwidth Profiles None have been defined
Dedicated • Timing over Packet
Tools • Security (MACSec)
Developed By • IEEE • IETF
• Multiple competing SDOs • Multiple competing SDOs
Ethernet vs. MPLS-TP in the Access Slide 11
13. Comparison Criteria
1. Fault Management Functionality
2. Performance Management
Functionality
3. Automatic Protection Switching
Mechanisms
4. Quality of Service Mechanisms Each will be scored for:
5. Traffic – Handling Diverse Client Types Suitability: 2 points
6. Timing – High Accuracy Time and Coverage: 4 points
Frequency Distribution Maturity: 4 points
7. Integration with Surrounding
Networks
8. CapEx
9. OpEx
10. Security
Ethernet vs. MPLS-TP in the Access Slide 13
14. Fault Management Functionality:
The Arguments
The Need: Access networks require strong FM capabilities to minimize down-time
Ethernet
• Two OAM mechanisms (Y.1731/CFM and EFM)
• Unique source address, particularly amenable to trace-back functionality
• Q-in-Q is not true client-server, but this is covered up by MEL
• Y.1731 is full-featured – comprehensive set of FM TLVs
• EFM is more limited, but adds dying gasp critical for CPEs
• Interop issues of both OAMs have finally been resolved; remaining details
are resolved via implementation agreements (e.g., MEF-30)
MPLS-TP
• Had basic heartbeats (BFD) and diagnostics (LSP-ping)
• The IETF designed MPLS-TP FM based on the Generic Access Channel and:
• BFD for CC
• LSP-ping for on-demand diagnostics
• New frame formats to fulfill specific requirements
Ethernet vs. MPLS-TP in the Access Slide 14
15. Fault Management Functionality:
The Verdict
Ethernet MPLS-TP
No true address, requires
Highly suited (SA)
Suitability extra work
2 Points 1 Point
Y.1731 is full featured, EFM MPLS-TP FM similar to CFM
Coverage fulfills its requirements but missing dying gasp
4 Points 3 Points
Y.1731 and EFM are Some MPLS-TP features are
interoperable and widely seeing initial trials
Maturity deployed
4 Points 1 Point
Total 10 Points 5 Points
Ethernet vs. MPLS-TP in the Access Slide 15
16. Performance Management Functionality:
The Arguments
The Need: Useful tool for maintenance and diagnostics of the access network
Ethernet
• The ITU Y.1731 supports PM (loss, delay, PDV, …) using a request-response
model; IEEE 802.1ag does not
• Y.1731 is used as the basis for commissioning procedures (Y.1564)
• Widespread vendor interoperability has been demonstrated
MPLS-TP
• RFCs 6374, 6375 define a set of PM functions based on the GA Channel
• These functions were designed to be HW friendly, yet flexible
- Support byte or packet counters
- 1588 or NTP-style timestamps
- Traffic-counters or synthetic loss
• Implementations have yet to be announced
Ethernet vs. MPLS-TP in the Access Slide 16
17. Performance Management Functionality:
The Verdict
Ethernet MPLS-TP
Neither protocol has an inherent advantage or disadvantage
Suitability
2 Points 2 Points
Supports all features (may
Supports all features
Coverage be more flexible)
4 Points 4 Points
MPLS PM is not (widely)
Y.1731 is finally interoperable
Maturity implemented
4 Points 0 Points
Total 10 Points 6 Points
Ethernet vs. MPLS-TP in the Access Slide 17
18. Automatic Protection Switching:
The Arguments
The Need: APS is a complex subject, requiring careful protocol work and proper
configuration. Solutions required for both linear and ring protection
Ethernet
• Ethernet has a particular problem with rings
• There are many open loop ring protection (e.g., G.8032) but these are not
compatible with QoS mechanisms
MPLS-TP
• MPLS in the core exploits Fast ReRoute (RFC 4090) instead of APS
• But FRR requires rich interconnection and so is usually not applicable
to access networks
• The IETF has standardized RFC 6378 for MPLS-TP linear protection
• There are also proposals for ring protection
Ethernet vs. MPLS-TP in the Access Slide 18
19. Automatic Protection Switching:
The Verdict
Ethernet MPLS-TP
No particular strengths or
Not suitable for ring protection
Suitability weaknesses
0 Points 2 Points
G.8031/G.8032 fulfill current RFC 6378 (linear protection);
Coverage requirements no ring protection RFC yet
3 Points 2 Points
G.8031/G.8032 have been MPLS-TP only partially
extensively debugged and finalized and not yet
Maturity updated deployed
4 Points 1 Point
Total 7 Points 6 Points
Ethernet vs. MPLS-TP in the Access Slide 19
20. Quality of Service:
The Arguments
Two types of QoS need to be considered to manipulate traffic:
1. Hard QoS (engineering): Connection Admission Control, Resource Reservation
2. Soft QoS (conditioning): Priority marking, discard eligibility, queuing, bucketing
Ethernet
• PBB-TE (PBT) defines hard QoS, but is not widely implemented
• P-bits for prioritization marking; discard eligibility marking in S-VLANs
• MEF’s BW profile defines a bucketing algorithm
• Ethernet headers are self-describing – support Traffic Awareness
MPLS-TP
• MPLS-TE supports resource reservation but traffic engineering may not
be relevant for access networks
• Traffic Class (and L-LSPs) support DiffServ prioritization application
awareness – MPLS packets are not self-describing
• DPI is required for Traffic Awareness
Ethernet vs. MPLS-TP in the Access Slide 20
21. Quality of Service:
The Verdict
Ethernet MPLS-TP
Does not define for (bucket-
Supports all QoS types
Suitability based) traffic conditioning
2 Points 1 Point
MEF standards have been Without bucketing,
Coverage proven MPLS is at a disadvantage
4 Points 3 Points
BW profiles are standardized;
MPLS-TP – nothing special
Maturity certification programs
4 Points 0 Points
Total 10 Points 4 Points
Ethernet vs. MPLS-TP in the Access Slide 21
22. Traffic types:
The Arguments
The Need: No transport protocol is useful if it can’t transport the required client
traffic
Ethernet
• Differentiates traffic via Ethertype marking or LLC
• Can directly carry IPv4, IPv6, MPLS, Ethernet, fiber channel, and low-
rate TDM (MEF-8)
• Does not directly carry other legacy traffic types (e.g., ATM, frame relay)
• Can be carried indirectly via PHP’ed MPLS PWs
MPLS-TP
• Can carry IPv4, IPv6, MPLS, and PWs (and through which Ethernet, Fiber
Channel and all legacy types)
• Defining a new PW type requires IETF consensus but the new packet-PW
provides more freedom
Neither is universal but existing mechanisms can be extended to cover
new cases
Ethernet vs. MPLS-TP in the Access Slide 22
23. Traffic types:
The Verdict
Ethernet MPLS-TP
Supports arbitrary clients via Supports arbitrary clients via
Suitability Ethertypes PWs
2 Points 2 Points
Does not support all legacy Supports most traffic types
Coverage traffic types (ATM, FR) via PWs
2 Points 3 Points
Ethertypes have been widely PWs have been widely
Maturity deployed deployed
4 Points 4 Points
Total 8 Points 9 Points
Ethernet vs. MPLS-TP in the Access Slide 23
24. Timing:
The Arguments
The Need: Distribution of highly accurate timing (frequency and Time of Day)
is crucial for some access network applications, notably cellular backhaul
Ethernet
Two protocols have become standard for this purpose:
1. Synchronous Ethernet (SyncE) is Ethernet-specific
2. IEEE 1588-2008 (defined for Ethernet and UDP/IP) for Timing over
Packet; on-path support elements (Boundary Clocks or Transparent
Clocks) have only been defined for Ethernet
MPLS-TP
• MPLS does not define a physical layer
• The IETF TICTOC WG is presently working on 1588oMPLS
Ethernet vs. MPLS-TP in the Access Slide 24
25. Timing:
The Verdict
Ethernet MPLS-TP
May be able to support
Supports ToP and defines a
1588 but there can be no
Suitability physical layer to support SyncE
SyncMPLS
2 Points 1 Point
Meets all requirements with 1588oMPLS to support ToP
Coverage SyncE, 1588, BC, TC may be coming
4 Points 1 Point
ITU-T has defined profile(s) for MPLS presently has no
Maturity 1588 use timing support
4 Points 0 Points
Total 10 Points 2 Points
Ethernet vs. MPLS-TP in the Access Slide 25
26. Integration:
The Arguments
The access network needs to integrate with the core and customer networks.
Cost and complexity will be minimized by a smooth hand-off, i.e., access
protocol compatibility with other network protocols
• Customer networks may have Ethernet or TDM interfaces (IP over Ethernet,
Ethernet over TDM, Ethernet over SDH)
• Core networks are usually MPLS (IP over MPLS, MPLS over Ethernet, MPLS
over SDH)
Ethernet
• Ethernet in the access is a perfect match for customer networks
• Can not seamlessly interface with MPLS core
MPLS-TP
• A reasonable match for customer network protocols since they can be
tunneled over MPLS
• Reuses existing MPLS standards thus maximizing compatibility
Ethernet vs. MPLS-TP in the Access Slide 26
27. Integration:
The Verdict
Ethernet MPLS-TP
Best match for core
Perfect match for customer
network, but not for
Suitability network, but not for core
customer
1 Point 1 Point
Does not require GW for
Ethernet Q-in-Q and MAC-in-
forwarding to core, but
MAC perfect customer hand-
Coverage control protocols may not
off
interconnect
3 Points 2 Points
Ethernet Q-in-Q presently Seamless MPLS still in its
Maturity widely deployed infancy
4 Points 1 Point
Total 8 Points 4 Points
Ethernet vs. MPLS-TP in the Access Slide 27
28. CapEx:
The Arguments
The Need: Access network providers need to keep their costs down; due to the
large number of Network Elements, access networks are CapEx-sensitive
Ethernet
• Ethernet switching fabrics are inherently non-scalable (long global
addresses can’t be aggregated)
• Due to popularity, Ethernet switches are inexpensive (high volumes, large
R&D investment in cost reduction)
• However, carrier-grade Ethernet switches need extra functionality
• Ethernet supports CapEx-saving architectures (e.g., EPON)
MPLS-TP
• LSRs are complex and expensive – reducing the price of NEs (MPLS switch
instead of MPLS router) was the unstated motivation for MPLS-TP
• Pure MPLS NEs have simple forwarding engines and thus should be less
expensive than Ethernet switches, but still require Ethernet or SDH or
OTN interfaces
Ethernet vs. MPLS-TP in the Access Slide 28
29. CapEx:
The Verdict
Ethernet MPLS-TP
Inexpensive, but can not scale Allows for significant cost
Suitability forever reduction vs. full LSR
1 Point 2 Points
R&D and volumes have driven MPLS-TP-specific devices
Coverage down Ethernet CapEx can be low cost
4 Points 4 Points
Many trials are using LSRs;
MEF certification programs for
chip sets are starting to
Maturity carrier-grade Ethernet switches
come out to address
4 Points 2 Points
Total 9 Points 8 Points
Ethernet vs. MPLS-TP in the Access Slide 29
30. OpEx:
The Arguments
• OpEx considerations that are taken into account:
- Direct operations cost
- Staffing
- Minimizing “unchargeable” overhead
• Reduction of direct operations costs for networks with large number of NEs :
- Equipment must work reliably and be interoperate
- Minimum touch (auto-discovery, zero-touch configuration., etc.)
- Use of FM, Control Plane or Management Plane protocols
• Maintaining competent staff requires:
- Availability
- Training
- Employee retention
• Overhead minimization applies to:
- Per packet overhead
- OAM, CP/MP packets
Ethernet vs. MPLS-TP in the Access Slide 30
31. OpEx:
The Arguments (Cont’)
Ethernet
• Basic Ethernet is zero-touch by design but carrier-grade may add many
configuration parameters
• A large number of useful L2CPs (STP, ELMI, GVRP) but no universal CP
protocol
• In addition to equipment certification, MEF has initiated certification for
carrier Ethernet engineers
• Main Ethernet overhead is large, but tags add only a small delta
MPLS-TP
• Basic MPLS relies on IP routing protocols but TP is designed to be able to
function w/o CP
• GMPLS CP has been defined as an option
• Can operate without IP forwarding (eliminating IP logistics); CP and MP
can be carried in GACh (although not yet developed)
• Specific vendors have expert certifications but none specific to MPLS-TP
• Same look and feel as other transport networks to minimize retraining
• May leverage extensions to existing OSS
Ethernet vs. MPLS-TP in the Access Slide 31
32. OpEx:
The Verdict
Ethernet MPLS-TP
Metro Ethernets have been Designed to be
Suitability shown to be low OpEx inexpensively maintainable
2 Points 2 Points
Inelegant CP, available staff, Learned from previous
Coverage medium overhead efforts
4 Points 4 Points
Extensive operational
Extensive experience and
experience only partially
Maturity certification programs
applicable
4 Points 2 Points
Total 10 Points 8 Points
Ethernet vs. MPLS-TP in the Access Slide 32
33. Security:
The Arguments
The Need: Security is perhaps the most important telecomm issue today
• OAM, APS, QoS mechanisms are powerless to cope with Denial of Service
attacks
• Access network NEs are frequently physically unprotected, so:
1. Ports must be protected
2. Packets must be authenticated and integrity checked
3. Confidentiality mechanisms may be needed
4. MPs and CPs must be hard-state
Ethernet vs. MPLS-TP in the Access Slide 33
34. Security:
The Arguments (Cont’)
Ethernet
• Ethernet packets carry unique authenticatable source addresses
• MACsec and its 802.1X extensions define mechanisms that can be used to
protect carrier networks (although hop-by-hop security model may not
always be ideal)
MPLS-TP
• MPLS was designed for core networks (walled gardens) with the
assumption that there are no inside attacks
• Forwarding plane attacks based on lack of authentication/integrity
• Control plane attacks based on soft state of protocols
Ethernet vs. MPLS-TP in the Access Slide 34
35. Security:
The Verdict
Ethernet MPLS-TP
Has an authenticatable unique Has no source identifier and
Suitability SA uses soft-state CPs
2 Points 0 Points
MACsec and 802.1X, but may Little positive support (but it
Coverage need more does support attacks …)
3 Points 1 Point
MPLS community is not
MACsec is starting to appear in
addressing the TP security
Maturity standard chipsets
problem
2 Points 0 Points
Total 7 Points 1 Points
Ethernet vs. MPLS-TP in the Access Slide 35
36. Summary: Final Score
Suitability Coverage Maturity Total
Ethernet 16/20 35/40 38/40 89
MPLS-TP 14/20 27/40 11/40 52
Note: MPLS-TP lost
• 29 points due to lack of maturity
• 9 points due to lack of security
Caveats:
• Deployments have particular (non)requirements, but all 10
considerations were given equal weight
• Some coverage and all maturity scores will change over time
Ethernet vs. MPLS-TP in the Access Slide 36
37. For information about RAD’s
Ethernet Access solutions,
visit www.ethernetaccess.com
www.rad.com
Ethernet vs. MPLS-TP in the Access Slide 37