comparison of on-path support to other methods for ... · unaware networks • no processing of the...

25
Power Matters Comparison of On-path Support to Other Methods for Precision Timing Protocol Adam Wertheimer Microsemi Corporation +1-613-270-7235 [email protected] © 2013 Microsemi Corporation 1

Upload: others

Post on 12-Jun-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters

Comparison of On-path Support to Other Methods for Precision Timing Protocol Adam Wertheimer Microsemi Corporation +1-613-270-7235 [email protected]

© 2013 Microsemi Corporation 1

Page 2: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Delivery of Frequency via Synchronous Ethernet Methods for delivering phase

• Phase over Unaware networks • Phase over Partially aware network • Phase over Aware networks Comparison of results using different methods

Introduction

2 © 2013 Microsemi Corporation

Page 3: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Fully approved in ITU-T G.8262 Advantages

• Synchronous Ethernet extends the SONET/SDH timing model to Ethernet

• Meets all existing frequency requirements via the bit rate of the Ethernet physical layer – Independent of packets and loading

Disadvantages • Need to upgrade Ethernet equipment in the Ethernet packet chain to

support SyncE • Need unbroken chain of SyncE equipment from frequency source to

end application – May use SONET or PDH to add timing to Ethernet at some intermediate

point in the network (i.e. at egress from SONET over packet network to Ethernet network)

Use case – SyncE for frequency distribution

3 © 2013 Microsemi Corporation

Page 4: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

SyncE Syntonization

4 © 2013 Microsemi Corporation

BITS/SSU PRS/PRC

G.812 Type I SSU

G.811 PRC

G.8262 EEC

Option 1 End

Application

End Application

Page 5: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Transition from Frequency to Frequency and Phase

5 © 2013 Microsemi Corporation

Page 6: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Aware • Addition of Boundary clock at each node in the network

– According to ITU model current under study • Split up the network in to smaller pieces • Needed for end-to-end time of day performance

Terminology: Aware networks

6

From IEEE Std 1588-2008 page 32

© 2013 Microsemi Corporation

Page 7: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary

Clocks Partially Aware

• Some Boundary Clocks in the network but not at every node • May be needed for existing networks during transition • May allow phase transfer without upgrading all network elements in

network Aware Networks

• All nodes in the network have Boundary Clocks

Terminology: Network Types

© 2013 Microsemi Corporation 7

Page 8: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Network Types

© 2013 Microsemi Corporation 8

Unaware

Partially Aware

Aware

Page 9: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Unaware No On-Path Support

9 © 2013 Microsemi Corporation

Telecom Slave /BC (IEEE1588)

PRS/PRC

PRTC

IEEE1588 Server

G.8272 PRTC

G.8273.1 PTM T-GM

G.812 Type I SSU

G.811 PRC

G.8262 EEC

Option 1

“G.82xx” PEC-B

BITS/SSU

Page 10: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Test Setup

10

1PPS

IP Traffic

Scope

sync, delay_resp delay_req

Grand Master

GPS

Ext. Ref

Network Emulator

5 or 10 GE Switches

1PPS

Slave

IP Traffic

© 2013 Microsemi Corporation

Page 11: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Results – 5 switch network

5 Gigabit Ethernet switch network Passed frequency mask for DS1 and E1 network

© 2013 Microsemi Corporation 11

Test Case Traffic Model Phase (30 min from start-up)

Minimum [ns]

Maximum [ns]

Absolute [ns]

TC12 - Static TM2 -92 5 92

TC13 – Square (50min from start)

TM2 -220 90 220

TC13 – Square (70min from start)

TM2 -180 102 180

TC14 - Ramp TM2 -140 30 140

TC15 - Outage 10 & 100 s.

TM2 -62 25 62

TC16 - Cong. 10 & 100 s.

TM2 -52 -5 52

TC17 – Re-route 3 Sw 200us.

TM2 -65 63 65

Page 12: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Results – 10 switch network

10 Gigabit Ethernet switch network Passed frequency mask for DS1 and E1 network

© 2013 Microsemi Corporation 12

Test Case Traffic Model Phase (30 min from start-up)

Minimum [ns]

Maximum [ns]

Absolute [ns]

TC12 - Static TM2 -220 +150 220

TC13 – Square (50min from start)

TM2 -200 +285 285

TC13 – Square (70min from start)

TM2 -330 +280 330

TC14 - Ramp TM2 165 +265 265

TC15 - Outage 10 & 100 s.

TM2 165 +232 232

TC16 - Cong. 10 & 100 s.

TM2 115 +260 260

TC17 – Re-route 3 Sw 200us.

TM2 190 +285 285

Page 13: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Network Types

© 2013 Microsemi Corporation 13

Unaware

Partially Aware

Aware

Page 14: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Currently under study in standards Too many network types and configurations Need unaware phase profile May be possible in a managed network

• Single carrier with careful engineering of link utilization and routing

Use case Phase over Partially aware networks

14 © 2013 Microsemi Corporation

Page 15: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Partially Aware Network Partial On-Path Support without SyncE Syntonization

15 © 2013 Microsemi Corporation

Telecom Slave /BC (IEEE1588)

PRS/PRC

PRTC

IEEE1588 Server

G.8272 PRTC

G.8273.1 PTM T-GM

G.812 Type I SSU

G.811 PRC

G.8262 EEC

Option 1

“G.82xx” PEC-B

BITS/SSU

BC (optional)

Page 16: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Test Setup

16

1PPS

IP Traffic IP Traffic

1PPS

Scope

sync, delay_resp delay_req

Grand Master Modified BC

GPS

Ext. Ref

Network Emulator

10 GE Switches

sync, delay_resp delay_req

Ext. Ref

Network Emulator

10 GE Switches IP Traffic

1PPS

Slave

IP Traffic

© 2013 Microsemi Corporation

64/16 64/16

Page 17: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Results - Frequency

Passed frequency mask for DS1 and E1 network

Test Case Traffic Model ITU-T G.8261 PEC Deployment Case 2

ITU-T PEC Deployment Case 1 / EEC Option 1 / ITU-T G.823 SEC

G.823 Traffic CES MTIE TDEV

BC: TC12 – TC17 TM2

Slave: TC12 – TC17 TM2

© 2013 Microsemi Corporation 17

Test Case Traffic Model

ITU-T G.8261 PEC Deployment Case 2 G.824

DS1

G.824 Traffic (TDEV) G.824 Traffic CES

BC: TC12 – TC17 TM2

Slave: TC12 – TC17 TM2

Page 18: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Results - Phase

Passed phase mask of 1 microsecond over the 20 switch network Network: 10 GE switches – modified BC – 10 GE switches Used combined PDV file with 7 test cases

© 2013 Microsemi Corporation 18

Test Case Traffic Model

Phase

Min Max Abs

BC: TC12 – TC17 TM2 -550 ns 334 ns 550 ns

Slave: TC12 – TC17* TM2 -939 ns 297 ns 939 ns

Page 19: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Partially Aware Network Partial On-Path Support with SyncE Syntonization

19 © 2013 Microsemi Corporation

Telecom Slave Clock (IEEE1588 & SyncE)

PRS/PRC

PRTC

IEEE1588 Server

G.8272 PRTC

G.8273.1 PTM T-GM

G.812 Type I SSU

G.811 PRC

G.8262 EEC

Option 1

“G.82xx” PEC-B

BITS/SSU

BC (optional)

Page 20: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Network Types

© 2013 Microsemi Corporation 20

Unaware

Partially Aware

Aware

Page 21: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Aware Network Full On-Path Support without SyncE Syntonization

21 © 2013 Microsemi Corporation

G.8272 PRTC

G.8273.1 PTM T-GM

G.8273.2 T-BC

G.811 PRC

G.8262 EEC

Option 1

G.8273.4 T-TSC

PRTC

IEEE1588 Server Telecom Slave Clock

(IEEE1588)

PRS/PRC BITS/SSU

Page 22: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Test Setup

Master, 10 Boundary Clock nodes, Slave Expect phase alignment of 150-200 ns

• Based on noise accumulation along the chain at 16/16

© 2013 Microsemi Corporation 22 22

1588 CLK & PPS (measure)

T-BC #10

S M

T-BC #9

S M

T-BC #8

S M

T-BC #7

S M

T-BC #6

S M

T-BC #5

S M

T-BC #4

S M

T-BC #3

S M

T-BC #2

S M

T-BC #1

S M

T-GM #0

M

PRTC (GPS or Rb)

T-TSC #11

S

1000 Mbps Copper

IEEE 1588 Time Domain

SyncE Frequency Domain

1588 CLK & PPS (measure)

Page 23: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Aware Network Full On-Path Support with SyncE Syntonization

23 © 2013 Microsemi Corporation

G.8272 PRTC

G.8273.1 PTM T-GM

G.8273.2 T-BC

G.811 PRC

G.8262 EEC

Option 1

G.8273.4 T-TSC

PRTC

IEEE1588 Server Telecom Slave Clock

(IEEE1588 & SyncE)

PRS/PRC BITS/SSU

Page 24: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Test Setup/Results

Master, 10 Boundary Clock nodes, Slave • SyncE distribution reference chain Phase (at slave)

• FFO < +/- 3 ppb • MTIE < 30 ns • Time Error < 70 ns

– Note: Packet rate of 16/16

© 2013 Microsemi Corporation 24 24

1588 CLK & PPS (measure)

T-BC #10

S M

T-BC #9

S M

T-BC #8

S M

T-BC #7

S M

T-BC #6

S M

T-BC #5

S M

T-BC #4

S M

T-BC #3

S M

T-BC #2

S M

T-BC #1

S M

T-GM #0

M

PRTC (GPS or Rb)

T-TSC #11

S

1000 Mbps Copper

IEEE 1588 Time Domain PRC/ PRS (Rb) SyncE Frequency Domain

SyncE Clock

(measure)

1588 CLK & PPS (measure)

SyncE Clock

(measure)

Generic Switch/Router

Clock Filtering

Ethernet/IPSyncE/SONET/SDH

Line Cards

Ethernet/IPSyncE/SONET/SDH

Line Cards

Telecom-BCG.8273.2

L2/L3 Sync Network

L1 SyncNewtork

SyncE/SONET/SDHLine Card

Recovered Clock

Ethernet/IPLine Card

With MAC/PHY TImestamping

Ingress Packets /

Timestamps

EECG.8262

Ethernet/IPSyncE/SONET/SDH

Line Cards

L1 SyncNetwork

L2/L3 Sync Network1588 Clock

1588 PPS

SyncE Clock

L1 SyncNetwork

L2/L3 Sync Network

FilteredSyncEClock

1588 ToD

Page 25: Comparison of On-path Support to Other Methods for ... · Unaware networks • No processing of the PTP packets at intermediate nodes by Boundary Clocks Partially Aware • Some Boundary

Power Matters.

Summary

© 2013 Microsemi Corporation 25

Phase transfer results for various networks as shown in this presentation The use of aware network with SyncE support give the best

performance The use of SyncE provides improvement in the partially

aware case SyncE and IEEE-1588 together gives the best performance

Phase transfer (ns) Partially Aware Aware

No SyncE Support Good Better

With SyncE Support Good Best