CX5003 1588 Slave Clock App Note v7
CX5003 1588 Slave Clock App Note v7
CX5003 1588 Slave Clock App Note v7
CX5003
CONTENTS
1
1.1
Equipment required
1.2
Common Stratum-1 (or better) GPS reference clock with frequency output
(10MHz, 2MHz) and phase reference (1pps).
Accessories and cabling: BNC cables with T-splitters, RJ45 Ethernet cables,
SFPs, matching single-mode or multi-mode fibers
Fig 1. - ITU-T G.8261 Appendix VI Setup with a Test Equipment acting as the 10 switch network
1.3
Fig.2 Test Setup for running ITU-T G.8261 Performance tests with Paragon-x
The following is a step-by-step guide to setting up the Paragon-x to run G.8261 profile tests:
10GbE Port 2
Status Display
100MbE/1GbE
Port 1
100MbE/1GbE
Port 2
PC Controller Port
The front panel of the Paragon-x provides different connections that can be used for testing:
1. 100Mbit/s Electrical or optical (via SFP) ethernet
2. 1Gbit/s Electrical or optical (via SFP) ethernet with option 110 fitted
3. 10Gbit/s Optical input (XFP or SFP+) with option 111 fitted
Important Tip: When using optical ports it is important to remember that both SFP modules
must be inserted into the Paragon-x in order to select the optical option in the Physical
Settings window.
1. Connect the 1588v2 Master Clock (shown in Fig.2) to Port 1 of the Paragon-x.
2. Connect the 1588v2 Slave Clock (shown in Fig. 2) to Port 2 of the Paragon-x
Note: Both connections must be at the same rate (and same media Electrical or Optical). If
the Master and Slave clocks in your test setup are set to run at different rates, add a switch to
the test set up.
Connect the common synchronisation reference to the relevant Reference Input connector.
Connect the common 1pps reference or master clock 1pps reference output to the
appropriate Paragon-x 1pps reference input 12 or 16 in fig. 4, depending on cable and
connector type appropriate to common reference or master clock used.
5. Press Setup Interface. On Ethernet tab, select Thru Mode and Ethernet interface
type/rate. Note if using SFP/XFP devices, one must be present in both ports for either
to be selectable.
6. Select the references tab and set reference clock source according to rear panel
reference input used. Set the 1pps reference selection according to the 1pps reference
input used to connect the common frequency reference or master clock 1pps reference
output to the Paragon-x.
The G.8261 Test Case PDV profiles are run on the SYNC messages in the forward direction
and the DELAY_REQ messages in the reverse direction.
The Paragon-x uses a tool called Flow Wizard to select the 1588v2 flows and set the filters
to target these flows.
10
1. Click on the
below)
2. Click on
11
5. Flow Wizard will automatically group flows that have the same value for each of the
fields displayed. To show the individual messages, Right click on Pink row above and
then click Expand row to show individual message types.
6. Select flows of interest, multiple flows may be selected by using either the Shift and left
mouse button or Ctrl and left mouse button. For G.8261 testing, select the SYNC and
DEL_REQ flows.).
7.
Click
8.
Click
9.
The Flow Filter Window in the main Paragon-x GUI (shown below) will now show that
filters have been set.
12
Sync, Del_Req, Del_Resp should not exceed 128pps as specified in the ITU-T for
Telecom unicast applications
You can validate the 1588 messages between the Master and Slave as follows:
1. Select the Start Capture button to start measurement.
2. If desired, stop the measurement (after a pre-defined period for running the test), select
Stop Capture.
The header and timing table allows the user to quickly determine if there are any misordered, missing or repeated packets.
3. To do this, look at the top of the SequenceID column and check the soft LED icon.
Indicates no error
Indicates there is an error
13
14
15
4. Now you are ready to import a G.8261 test case profile. You will need to import a profile
for each direction. The G.8261 profile files available on the Calnex website are labeled to
identify their direction and packet rate, two examples are:
a. g8261_tc12_forward_SYNC_32pps
b. g8261_tc12_reverse_DREQ_32pps
Where: g8261_tc12 identifies the specific ITU-T G.8261 test to be performed, and
forward_SYNC identifies the direction and 1588v2 messages within the profile, and
32pps identifies the packet rate.
5. Import the FWD profile for Port 1 > Port 2 direction by selecting the Import button
within the Add Impairments and Delay window (note: if this selection is greyed out
ensure that the Variable Delay Insertion enable box is checked).
16
6. Once the import button has been clicked, browse your PC for the location of the FWD
profile you wish to load.
7. Click Open, the profile is now enabled for the Port 1 > Port 2 direction
8. Click on the Port 2 > Port 1 tab and repeat steps 7 through 9.
9. To start the overwrite click
Replay
Progress
The replayed profile for each direction is displayed. Each graph has a red vertical bar above
it; this bar shows the progress of the replay of the delay profile.
For G.8261 testing, there is a required amount of time that each test must run before
checking the results. Review Appendix 2 for further information.
17
Important Note: The G.8261 Test case must be started before the
measurements on the slave clock. There is a Stabilisation period allowed
in G.8261 (900seconds or 15minutes). The Calnex G.8261 Test Case
profiles have a stabilisation period of 15minutes at the beginning, so start
the profiles as shown above but only start the next step after 15 minutes.
18
1.4
For performance evaluation of a G.8261 profile, the recovered clock of the 1588v2 Slave
should be measured as follows:
a)
Frequency output (MTIE/TDEV measurement) if 1588v2 is being tested for
Frequency transfer
b)
1pps Time of Day output (ToD/1pps Error measurement) if 1588v2 is being
tested for Time/Phase transfer
Slave Clock
19
For the pinout, refer to the figure below; Pin 1 will receive the measured input while
the ground (or 0V signal) is on Pin 2. The other pins are not used.
NOTE: The 1pps measurement port (Top AUX connector) has a minimum
Voltage requirement of 1.65V
20
1. Verify that 1588v2 mode is still selected and E1/T1 Wander and 1pps Accuracy
measurements are selected
21
22
23
7. If necessary, it is possible to export the TIE data in .csv format and use Excel to
edit out time segments (eg, results obtained during settling time). The edited TIE
data can then be loaded back into the Paragon-x GUI and re-analysed using the
Wander Analysis Tool as above.
When editing such a file, its vital to retain the first line of data, shown as line 21
in the example below.
24
25
1.5
If the slave clock output fails the MTIE/TDEV or 1pps measurements,the next step should be
to try and isolate the cause of the failure. This can be challenging for G.8261 tests as there
can be a number of contributing factors to why a test has failed. Here are some areas to
investigate before re-running the test. This is not an exhaustive list, and is provided for
guidance only:
1. Noise Floor Delay Raised during Peak Traffic During periods of high traffic,
the noise floor may raise and may increase the likelihood of test failure.
a. Adjusting Delay Floor It May be possible to adjust the Delay Floor
using the Paragons PDV Editor function. Appendix 1 describes the
PDV Editor in greater detail.
2. Check Quality of Clock Oscillator The quality of the clock oscillator (crystal)
should be minimum carrier-grade quality.
3. Check the Line Rate Youre running the test with a profile set for 1pps, but the
actual line rate is higher, this will result in a test failure.
4. Algorithm Working as Expected? Check your test parameters and verify
correct operation.
26
2.1
For live network capture the Paragon unit should always be placed next to the 1588v2 slave
as shown below:
2.2
Connect Paragon-x and proceed up to end of
selecting flows and filters section above.
Configure Capture: IEEE 1588
If required, the 1588 Settings window allows additional settings for 1588v2 capture.
1. Select
27
2. Select either 1 step (Sync only) or 2 step (Sync + Follow_Up) clock mode
3. Tick the box if the CorrectionField value contained in the Sync message is to be
included in the calculation of the captured PDV. This is typically used when
testing a 1588v2 Transparent Clock.
4. Enter the 1588 Packet rate This is used if captured data is exported to
Symmetricom TimeMonitor
5. ESMC Rx Monitoring - The instrument will capture ESMC messages in one
direction when in 1588 mode.
28
1. Select
directional information for each message will be displayed in the table and the graph
will show the inter-packet arrival time
2. The table continues to update in real-time with the captured bi-directional 1588v2
messages.
3. Select
4.
By default the Paragon shows all the fields in a 1588 message header, it is possible
to reduce the number of fields displayed to those of interest by selecting the 1588
Column sorter that can be found under Data in the top menu.
5.
29
2.4
Once you have completed a capture of your live network data, the data can be imported,
edited and analysed either in the field or in a laboratory environment.
By identifying the cause of the failure, operators and clock manufacturers can reengineer their network or equipment to prevent similar failures in the future. There are
two strategies that operators and network equipment manufacturers can use to reengineer their equipment to better cope with these network events:
1. Re-calibrate or re-engineer their clock equipment to better cope with the PDV
profiles that causes a clock recovery failure.
2. Re-engineer the network such that these catastrophic PDV patterns do not
appear again in the network. Operators can do the following depending on the
cause of the PDV pattern:
o
Re-route the PTP traffic to a less congested path (forward and reverse
direction).
Reduce the number of nodes between the Master and Slave by rerouting
the PTP traffic (forward and reverse direction).
30
31
Description
Sync PDV
Slave Wander
Pdelay_Req PDV
Pdelay_Resp PDV
Follow_up PDV
Delay_Resp RTD
Delay_Req PDV
Asymmetry PDV
Displays the Inter-packet arrival
time for a specific message against
Time.
Sync
Delay_Req
Pdelay_Req
Pdelay_Resp
Follow_Up
Delay_Resp
Displays the Inter-packet arrival
time against packet number
32
33
34
Follow Up PDV
The Paragon graphs the variation in arrival time of the Follow_Up message with
respect to the Sync message.
35
Slave Wander
The Paragon extracts the embedded timestamp within the Delay_Req message and
compares it to the Master Reference, the variation is then graphed to provide the
Slave Wander output as shown below.
36
37
Delay_Req PDV
The Paragon graphs the variation between the launch time (arrival time at the
Paragon) of the Delay_Req message and the embedded timestamp, t4, in the
Delay_Resp message.
38
39
Asymmetry PDV
The Paragon graphs the variation in the Master -> Slave with respect to the Slave ->
Master delay.
40
41
3.1
The packet distribution function allows the analysis of a delay data set from a capture to be
analysed and presented graphically as a Probability Density Function. The data to be
analysed must be present in the GUI of the instrument either from a capture or from loading
in a previous capture of profile. The tool is accessed in the following manner:
1. Once a capture has been loaded select the Graph menu item from the Paragon
main menu
2. Select the Plot PDF (Packet Distribution Function/Histogram) selection on the
Paragon application menu bar as shown below.
3. This will bring up the graphical display window for the data in the instrument as
shown in the figure below.
42
Once the graph has been opened, other captures can be analysed by repeating the selection
of Graph -> Open Packet Delay Distribution Graph. This may be done with the same data
or different data following a separate capture or profile load.
Once a Delay Distribution window is open it may be reused to analyze other data sets directly
without having to load the data into the instrument first:
1. On the Delay Distribution window click File -> Open then navigating to the results file
using the standard Windows Explorer interface.
2. The graph may be printed using the File -> Print selection and the format of the
printing may be seen using File -> Print Preview these controls are located in the
tool bar of the Delay Distribution window
43
3. On the right hand side of the Delay Distribution window there are zoom controls, X
Zoom and Y Zoom to allow parts of the graph to be viewed in greater or less detail.
These zoom in and out controls increment or decrement the area in small steps.
There are 2 additional commands located on the slider bars to allow faster zoom out
as shown below.
Fast
zoom
buttons
The analysis tool sorts the data set into 1000 buckets covering the range of values of the
data.
The OK button on the Delay Distribution window dismisses the graph window.
44
It is intended to aid experimentation to establish limits of operation and margin testing and to
allow network events to be combined to produce a composite profile which will speed up
testing.
The tool will operate on all profiles captured by Paragon when in IEEE 1588 or Services
modes of operation, and the supplied G.8261 test case profiles.
4. Edit the profile by placing the cursor on the graph display and pressing the left mouse
button. This will display the menu as shown below.
45
5. The editing functions that can be carried out from this menu are:
a. Remove this function allows part of the timing profile to be cut out and
disposed of or to retain part of the profile and dispose of the rest.
One use of this feature is to extract a short part of the profile to allow it to be
replayed to determine if it is a stressful section of a replay which causes a
slave clock failure.
First is to use the mouse. Place the cursor at the start of the section of
interest then left button down and drag the mouse to the end of the section of
interest. This will highlight the area.
The second way is to use the markers. These are set using the Markers
buttons shown on the screen shot below.
46
There are 2 methods of placing the markers. Using the mouse and using the
location. In each case both markers should be set. These are marked on
the graph as 1 and 2 as shown below.
47
To use the mouse, click the Markers button 1 or 2, a menu will appear,
select Set marker using mouse then place the cursor at the desired position
and left click. The marker will appear as shown in the previous screen shot.
Once the area of interest is selected the left mouse should be clicked and
then Extract and either Highlighted or Non-Highlighted selected.
If Highlighted is selected then only the area highlighted is retained and the
rest of the profile disposed of.
All where the overall profile is repeated the selected number of times. When
the All selection is made the following pop-up screen is displayed to allow
the repeat number to be set.
48
Original Profile
The other two sub-selections allow parts of the profile to be repeated within
the overall capture.
The selections allow the marking of the sections to be performed either using
the mouse or using markers as introduced in the Extract section.
Original Profile
c.
Copy and Paste. These are two entries in the menu and allow parts of the
profile to be copied then inserted into a later part of the profile.
Highlighted where the section identified by using the mouse click and drag
is copied
49
Replace highlighted allows the data in the copy buffer to replace the
highlighted area.
Replace between markers allows the data in the copy buffer to replace the
data between the two markers. Operation of the markers is explained in the
Extract section.
Insert before start inserts the data in the copy buffer before the start of the
data.
Insert after end inserts the data in the copy buffer after the end of the
currently loaded data.
Insert at marker inserts the data in the copy buffer at the point marked by
the appropriate marker. Operation of the markers is explained in the Extract
section.
50
51
52
f.
For a clearer view of the data it is recommended that the graph be viewed in
scatter graph mode rather than line mode.
The scatter graph mode is selected using the Select Graph -> Graph
Format -> Dot controls located at the right hand side of the graph display.
53
To define a band the Markers are used. Note that they set the band in the
Y-axis as shown below.
54
This is the percentage of all packets in the profile which will be concentrated
into the band defined by the two markers.
The packets selected to be moved into the band are picked at random and
given a randomised value within the band set.
The result will be as shown below where 50% of the packets have been
concentrated into the define band.
It is possible to apply the Banding over a section of the profile by highlighting the X
axis range over which the Depletion or Concentration is required.
For example the screenshot below shows the markers defining the band to be
adjusted and the highlighting to define an X axis section to be manipulated.
55
g. Noise floor Density allows the profile to be adjusted to change the density
of packets to be found in the noise floor i.e. lucky packets.
The band of interest is defined between the packet with the minimum delay
and one of the markers. Based on the requested %, the number of desired
packets within the noise band is then calculated.
If there are less packets than requested, packets out with the band are
linearly selected and reassigned a new random value in between the marker
and the minimum profile value.
If there are more packets in the band than requested then packets within the
band are linearly selected and offset by the difference between the marker
and the minimum profile value.
Example 1: Making an adjustment across all of the packets the user needs
to define the top of the band using the marker, Marker 2 in this case. This is
shown below.
56
The Banding and the market are selected as shown below. Since there is
no highlighting the only choice is All i.e. across the whole set of packets.
The following shows the effect when 80% of the packets are moved into the
band.
57
Example 2: Where the manipulation is applied over part of the packet set
using the highlighting feature.
h.
Undo Last allows the user to return to the data set prior to the last
operation.
Note that several windows may be opened to allow cutting and pasting of
parts of several profiles to provide a final composite data set for replay. This
is invoked by using the File -> Import selection from the application toolbar
menu.
59
Two traffic models have been defined in G.8261, identified as Network Traffic Model
1 and Network Traffic Model 2. A synopsis of each model is provided below:
Maximum size packets will occur in bursts lasting between 0.1s and 3s.
Maximum size packets will occur in bursts lasting between 0.1s and 3s.
Network disturbance load with 80% for the forward direction (Server to
Client) and 20%in the reverse direction (Client to Server) for 1 hour.
The test measurements should start after the clock recovery is in a stable condition.
Guidance on stabilization period is provided in appendix II of G.8261 and should be a
minimum of 900s. The disturbance background traffic to load the network must use
Network Traffic Model 2 as defined below:
60
The packets to load the network must use Network Traffic Model 1 (Test
Case 13a) as defined in ITU-T G.8261.
In the forward direction: Start with network disturbance load at 80% for 1
hour, drop to 20% for an hour, increase back to 80% for an hour, drop back
to 20% for an hour, increase back to 80% for an hour, drop back to 20% for
an hour. Simultaneously, in the reverse direction: Start with network
disturbance load at 50% for 1.5 hours, drop to 10% for an hour, increase
back to 50% for an hour, drop back to 10% for an hour, increase back to 50%
for an hour, drop back to 10% for 0.5 hour (see Figure VI.11/G.8261). Repeat
the test using the Network Traffic Model 2 (Test Case 13b) as defined above
to load the Network
61
The packets to load the network must use Network Traffic Model 1 (test case
14a) as defined above
In the forward direction: Vary network disturbance load smoothly from 20% to
80% and back over a 24-hour period. Simultaneously, in the reverse
direction: Vary network disturbance load smoothly from 10% to 55% and
back over a 24-hour period (see PDV Graphs below)
62
The packets to load the network must use Network Traffic Model 1 as
defined in VI.2.1
Start with 40% of network disturbance load in the forward direction and 30%
load in the reverse direction. After a stabilization period according to
appendix II, remove network connection for 10s, then restore. Allow a
stabilization period according to appendix II for the clock recovery process to
stabilize. Repeat with network interruptions of 100s.
63
Repeat the test using the Network Traffic Model 2 as defined in VI.2.2 to load
the network
The packets to load the network must use Network Traffic Model 1 as
defined in VI.2.1
Start with 40% of network disturbance load in the forward direction and 30%
load in the reverse direction. After a stabilization period according to
appendix II, increase network disturbance load to 100% in both directions,
(inducing severe delays and packet loss) for 10s, then restore. Allow a
stabilization period according to appendix II for the clock recovery process to
stabilize. Repeat with a congestion period of 100s.
64
Repeat the test using the Network Traffic Model 2 as defined in VI.2.2 to load
the network
Change the number of switches between the DUTs, causing a step change
in packet network delay.
o
The packets to load the network must use Network Traffic Model 1
as defined in VI.2.1
66
67
For more information on the Calnex Paragon-x, and to take advantage of Calnexs extensive experience in
synchronisation and packet testing technologies, please contact Calnex Solutions on +44 (0) 1506 671 416 or email:
info@calnexsol.com
www.calnexsol.com
This information is subject to change without notice
Calnex Solutions Ltd, 2012
CX5003 v7 October 2012
68