UMTS Network Insight Device Capacity

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 28

UMTS Network Insight Training

-- Device Capacity Analysis


Flow Control overview
① RNC: the software of each RNC board monitors
the system resource usage, and will starts flow
control functions when the usage is high

② UU:RNC performs CAPS (Call Attempts Per


Second) control, PCH congestion control and
FACH congestion control.

③ IU: When signaling bandwidth limited, RNC


performs flow control on initial UE messages to
reduce the signaling traffic.

④ Iub: Restrict transmission rates when there is


transmission congestion

Flow control is a protective measure for communications between the RNC and its peer equipment.
It restricts incoming traffic to:
Protect equipment from overload, thereby maintaining system stability.
Ensure that equipment can properly process services even under heavy traffic.
It restricts outgoing traffic to reduce the load on the peer equipment.
Flow Control -- RNC Units
Each RNC board monitors the following in real time to
keep track of resource consumption:
CPU usage: The CPU resources of a board determine the
processing capabilities of the board. All functions running
on the board use CPU resources.
Message block occupancy rate: Message blocks are
resources used to send and receive messages within the
RNC.
When the CPU usage or message block occupancy rate of a
board is high, the board processing capabilities may become
insufficient. And the board will triggers flow control to ensure
that basic functions can continue to run properly

Flow control based on message block occupancy rate and


CPU usage are independent. Related flow control functions
will be triggered when either the message block occupancy
rate or the CPU usage is excessively high.
Generally, it is rare to run out of message blocks.
If the CPU usage of an RNC unit reaches a preset threshold (configured by the SET CPUTHD command),
ALM-20256 CPU Overload is reported.
Flow Control -- filter window
Filter Window: To prevent frequent flow control triggered by CPU usage fluctuations, the system
also calculates the average CPU usage during a period of time that has just elapsed, and determines
whether to perform flow control based on this CPU usage. The CPU usage values used to calculate
the average CPU constitute a filter window.
Recommended Value: 10(SET FCCOMMPARA)

Fast judgement window: Number of CPU usage sampling times involved in fast judgment. The value of this
parameter must be smaller than or equal to half of the value of Filter window. Fast judgment window is applied to
control the flow in the case that the CPU usage is too high in a short time. After fast judgment window is applied,
the system compares the CPU usage in a short period of time before the current time with Critical threshold. If all
CPU usages in the fast judgment window reach or exceed Critical threshold, all active flow control mechanisms
start to work immediately
Recommended Value: 4(SET FCCOMMPARA)
Flow Control – SPU overload
CPUS or CP subsystem Flow Control
Impact on Controlling
Flow Control Function Flow Control Object Key Action
Services Command
Printing flow control Printing 1) Disable Printing,Debugging,Performance monitoring and Logging
Debugging flow control Debugging 2) Basic FC, and non-critical functions SET FCSW
Performance monitoring flow control Performance monitoring 3) Printing,Debugging,Performance monitoring THD= [50,60] SET
Logging flow control Logging 4) Logging THD= [80,90] FCCOMMPAR
No
A
1) Disable Resource audit
Resource audit flow control Resource audit SET
2) THD= [70,85]
FCCPUTHD
1) Disable
MR flow control MR function
2) THD= [50,60]
1) Discards paging messages order: SMS ->BE services, supplementary services, and location services->Real-
Paging control Paging messages time services
SET FCSW
2) Real-time services THD: [85, 90], BE,supplementary&location services THD: [75, 85], SMS THD: [70, 80]
1) AC restrictions
Access control based on the CPU
2) Domain-specific access control (DSAC)
usage or message block occupancy Users in AC0 to AC9
3) Intelligent AC control SET FCSW
rate
4) THD: [70, 80]
1) Rejects RRC connection requests with responding with RRC CONNECTION REJECT messages
2) rejects RRC connection requests with responding without RRC CONNECTION REJECT messages
RRC flow control RRC connection requests 3) THD-Inter-RAT cell reselection, IMSI detach procedure, registration, and incoming voice calls: [80, 90] SET
BE services and UE-originated voice calls:[75, 85] SHARETHD
SMS: [70, 80]
Yes
Flow control on signaling messages Some signaling messages 1) discards signaling messages(cell updates, handovers, and paging over the Iur interface)
over the Iur interface over the Iur interface 2)THD=[85, 95]
All messages over the Iur-g 1) discards signaling messages(cell updates, handovers, and paging over the Iur interface)
Flow control over the Iur-g interface
interface 2)THD=[85, 95]
All broadcast messages 1)reduce signaling traffic (By default disabled)
CBS flow control
delivered by the CBC 2)THD=[50, 60]
Some cell/URA update 1) discards all CBC broadcast messages
Cell update flow control SET FCSW
messages 2)[70, 85]
1) UEs CELL_FACH state, RNC does not trigger F2D
2) UEs CELL_DCH state, RNC does not trigger the DCCC procedure and HSUPA DCCC procedure
DCCC flow control DCCC procedure of the UE
3) No longer makes periodic attempts to trigger D2E or D2H transitions
4)THD=[75, 85](By default disabled)
SPU load sharing
SPU load sharing procedure 1.A CPUS receives an RRC connection request

2.Based on the current load, the CPUS determines whether to


perform load sharing. Load sharing is needed if the following
conditions are met:
The CPUS load is greater than or equal to
CtrlPlnSharingOutThd.
The remaining CPU processing capability of the CPUS in
the current second is insufficient to process the request

3.The CPUS forwards the request to another CPUS based on


the load information broadcast by the MPU
If Mean CPUs load of current subrack > Mean CPUs
load of whole RNC, and don’t need load sharing among
subracks.
If Mean CPUs load of current subrack < Mean CPUs
load of whole RNC, and don’t need load sharing among
subracks.
•If the CPUS load is 95% or higher, the CPUS discards load sharing requests for all services except those for
emergency calls
•Other CPUSs cannot forward load sharing requests to an overloaded CPUS
SPU load analysis
SPU load including two parts:
 One part of the process
associated with the call, this part can Sub rack resource Sub rack resource
MPU MPU
manage 3. SPU load balance manage
be sharing among the MPUs 4. Handling RRC Process
The other part of the Cell / NodeB
5. Cell/NodeB resource allocation request 8. Sending Uu RRC Msg
resource management and access User User
signaling signaling
control related, this part can not be 6. Resource allocation result
Cell/NodeB Cell/NodeB
sharing and need it sub SPU handle Manage 7. Sending IUB NBAP Msg(RL SETUP) Manage
which the NodeB is configurated SPU
2. Msg decoding SPU
Non Sharing Load:
Incoming RRC Req
IUB NBAP Msg,
Cell/NodeB resource allocation 1. Incoming RRC Req
request 9. Following RRC and NAS signaling Msg

Sharing Load:
RB
RB Setup
Measurement
Handover
•the CPUS load are included two part: Sharing Load and Non Sharing Load
•Subsystem CPU load > Load sharing THD, load sharing will start, but the Non Sharing Load is still handing by itself
SPU load sharing
load sharing among subracks

1. Each MPU collects the load information about all


CPUSs that are managed by the MPU.
2. Each MPU broadcasts the information to all CPUSs
3. If Mean CPUs load of current subrack < Mean
CPUs load of whole RNC, and don’t need load
sharing among subracks
SPU load analysis
Case 1 – SPU load
 Problem: 2_2_0 SPU load >70% ② Load sharing THD and RNC Average CPU Load:
“After software upgrade into R16 SPU Load should be balanced Para setting ok and no need expansion board
by itself like BSC or still can happen as below?”- From XL SPU Average Load RNC Level
RNCID RNC_SHARE_SWITCH CTRLPLNSHARINGOUTTHD CPU LOAD MEAN(%)

412 1 50 44.92

③ Sub System CPU Load: 2_2_0 Nonsharing load is high

SPUID SPU CPU Mean Real(%) Nonsharing Total(%)

1:0:7 SPUb 51.02 20.6


1:4:5 SPUb 51.08 30.88
2:2:0 SPUb 63.94 38.81
1:2:0 SPUb 51.28 19.1
2:8:7 SPUb 50.45 19.2
Analysis 1:8:0 SPUb 52.39 22.41
① NodeB Configuration Distribution by SPU: 1:4:4 SPUb 50.46 28.47
Unbalance: 2_2_0 with 7 NodeBs
Root cause: Subsystem 2_2_0 is configurated too many NodeB and
the Nonsharing load is high. When the Subsystem CPU load > 50%, it start
the load sharing, but the nonsharing load is still handled by 2_2_0 and the
CPU load is increasing, that why the load is closed to 70%

Solution: Rehoming NodeB to other subsystem which with lower


non sharing load.
Case 2 – SPU load
SPU Load Distribution Service for
Service for Service for Supplemen
IUR updating
100.00% Flow Call updating updating SMS tary service
downlin cells of Summary
Control paging cells of cells of low paging and LCS
90.00% k normal
high priority priority paging
priority
80.00% Subrack 0 7214 7 5867 109 12563 136733 162493
70.00% Subrack 1 63 33 1083 10 1807 187320 190316
60.00% Subrack 2 409 160 3523 54 7921 668584 2 680653
50.00% Subrack 3 138 10 2618 45 4708 203113 210632
40.00% Subrack 4 54 2933 56 6681 109444 1778 120946
30.00% Subrack 5 924 1108 32572 552 70860 1973658 6 2079680
Summary 8802 1318 48596 826 104540 3278852 1786 3444720
20.00%

10.00% CPU overload Distribution


300
0.00%
200
VS.XPU.CPULOAD.MAX VS.XPU.CPULOAD.MEAN
100

RNC_SHARE_SWI CTRLPLNSHARIN CPU LOAD 0


RNCID Busy Time
TCH GOUTTHD MEAN(%)
2015-07-29
162 1 30 55.77
19:00:00

 Finding: 1)the Average of CPULOAD>50%, 2) a lot of sub_XPU is overload, 3) the


CTRLPLNSHARINGOUTTHD=30%
 Impact: If CPU load is high, it impacts the service of the board.
 Action:1) 4 SPUb Board Expansion, 2)Rehoming NodeB, 3) CTRLPLNSHARINGOUTTHD=30%->50%
Flow Control – DPU overload
UP subsystem Flow Control

Impact on
Type Flow Control Function Flow Control Object Key Action Controlling Command
Services

Printing flow control Printing


1) Disable Printing, Debugging and Logging
High CPU usage
2) Basic FC, and non-critical functions
or message block Debugging flow control Debugging None SET FCSW
3) Printing, Debugging THD= [50,60]
occupancy rate
4) Logging THD= [80,90]
Logging flow control Logging

1) RNC lowers the rates of BE services


THD=[60,70]
1) DSP load>90%, it further lowers BE
High DSP CPU Flow control triggered by DSP
BE service rates service rates, abandons BE services and Yes None
usage CPU overload
CS services and prevents users in the
CELL_FACH state from accessing the
network
DPU load sharing
The UP load sharing procedure is as follows:

1. The CPUS receives UP resource allocation requests from users.


2. The CPUS selects a subrack based on the flow consumption:
 If the inter-subrack load=< the inter-subrack load THD InterSubRackTrafficThd,
selects the subrack where NodeB's controlling CPUS is located.
 Otherwise, selects the subrack where NodeB's interface board is located.
3. The CPUS determines whether load sharing is needed based on the UP load
on the selected subrack. If UP load on selected subrack >=
UserPlnCpuSharingOutThd, load sharing is needed.
4. The CPUS selects sharing subracks according to the following in equation:
UP load on a sharing subrack < UP load on the current subrack –
UserPlnCpuSharingOutOffset
5. The CPUS calculates the load sharing proportion based on the remaining
CPU processing capabilities of the current and sharing subracks. The CPUS
selects one of the subracks as the target subrack based on the sharing
proportion.
6. The CPUS selects a DSP subsystem from the target subrack as the target
DSP subsystem.
7. The target DSP subsystem allocates UP resources to the users.
DPU load analysis

•DSP is complete pool in the same MPU, so DSP load analysis base on MPU
Case 1 – DPU load
DPU CPU Load Distribution
DSP Usage Load Distribution 6.00%
70.00% 5.00%
4.00%
60.00% Subrack_1 3.00%

50.00% 2.00%
Subrack_2 1.00%
40.00% 0.00%
2015-07-22 2015-07-23 2015-07-24 2015-07-25 2015-07-26 2015-07-27 2015-07-28
30.00%

Sub Rack NodeB Num DPUe Num


20.00%
0 0 4
10.00% Subrack_0 1 68 4
0.00%
2 53 4
VS.DSP.UsagePeak VS.DSP.UsageAvg

CPU load: if the load is 70% -> suggest expansion,


USRPLN_PRORATION_ Average DSP
RNCID UserPlnCpuSharing
SHARED_ALO
Busy Time
Usage current is lower than 6%.
2015-07-27
DSP usage: if usage is more than 60%->suggest to do the
313 50 ON 29.32%
20:00:00
expansion. Current status is still lower than 50%

 Finding: DSP usage is unbalance


 Impact: It will increase SCU load when some DSP usage is high(>50%).
 Action:1) remove some DPU board from subrack_0 to other subrack. 2) rebalance some NodeB to subrack_0
Case 2 – DPU load
85
DSP.UsageAvg
DSP load for each MPU during one week
80
75
70
65
60
55
50
45
40
35
30
25
20
07/16/201…
07/16/201…
07/16/201…
07/16/201…
07/16/201…
07/17/201…
07/17/201…
07/17/201…
07/17/201…
07/17/201…
07/18/201…
07/18/201…
07/18/201…
07/18/201…
07/18/201…
07/19/201…
07/19/201…
07/19/201…
07/19/201…
07/19/201…
07/20/201…
07/20/201…
07/20/201…
07/20/201…
(空白)
Alarm analysis: BH, 18:00~ 24:00, all the MPU
occurred the alarm( by 07/16/2015)
Alarm MPU Occurred On (NT) Cleared On (NT)
22305 Subrack No.=4, Slot No.=16 07/16/2015 18:20:52 07/16/2015 23:02:52
22305 Subrack No.=0, Slot No.=3, 07/16/2015 18:22:24 07/16/2015 22:59:24
22305 Subrack No.=4, Slot No.=0, 07/16/2015 18:22:52 07/16/2015 23:01:42
22305 Subrack No.=5, Slot No.=2, 07/16/2015 18:22:56 07/16/2015 23:02:06 SET UALMTHD:
22305 Subrack No.=5, Slot No.=0, 07/16/2015 18:23:03 07/16/2015 23:01:13 USRPLNCPUALMTHD=70, USRPLNCPUCLRTHD=60
22305 Subrack No.=0, Slot No.=0, 07/16/2015 18:23:02 07/16/2015 22:57:52
22305 Subrack No.=1, Slot No.=3, 07/16/2015 18:23:26 07/16/2015 23:01:16 Impact on System:
22305 Subrack No.=1, Slot No.=0, 07/16/2015 18:26:54 07/16/2015 22:59:24
22305 Subrack No.=2, Slot No.=18 07/16/2015 18:48:54 07/16/2015 22:27:44
 Upload or Download rate decreases.Voice quality deteriorates.UEs
22305 Subrack No.=2, Slot No.=3, 07/16/2015 18:49:23 07/16/2015 22:36:43 fail to access the network
22305 Subrack No.=2, Slot No.=0, 07/16/2015 18:49:54 07/16/2015 22:22:52
22305
22305
Subrack No.=3, Slot No.=0,
Subrack No.=3, Slot No.=15
07/16/2015 18:49:54
07/16/2015 18:50:13
07/16/2015 22:25:44
07/16/2015 22:33:13
Solution: Rehoming NodeB or add new board
Flow Control – INT overload

Impact
Controlling
Type Flow Control Function Flow Control Object Key Action on
Command
Services

Printing flow control Printing


1) Disable Printing, Debugging and Logging
High CPU usage or message 2) Basic FC, and non-critical functions SET FCSW
Debugging flow control Debugging None
block occupancy rate 3) Printing, Debugging THD= [50,60] SET FCCOMMPARA
4) Logging THD= [80,90]
Logging flow control Logging

1) RRC connection requests


>INTFCTHRD=800,only processes RRC
connection requests from emergency calls
Flow control triggered by INT and high-priority services(incoming and
High CPU usage RRC connection requests SET TNSOFTPARA
overload on the control plane outgoing voice calls, inter-RAT cell
reselection, and registration)
2)exceeds 90%: discarding RRC
connection requests from all UEs.
Yes
1) the duration for buffering the data in
queue X > CONGTHDX(25ms Default
DPU reduce the transmission rate of the BE
Flow control triggered by Iub
Congestion in queues at the service
interface board overload on BE service rates
ports 2)the buffer length of the queue
the user plane
>DROPPKTTHDX(60ms default),
discarding the subsequent packets added to
queue
Flow Control – INT overload
Flow Control Triggered by INT Overload on the Control Plane

1. The interface board reports its CPU


usage to the MPU each second

2. MPU adjusts the maximum number of


session setup requests admitted by the
interface board

3. If the number of RRC connection


requests already admitted is larger than
the maximum number allowed, the RNC
only processes RRC connection
requests from emergency calls and
high-priority services
Flow Control – INT overload
Result of the backpressure-based flow control 1. The interface boards monitor the transmission buffers of the queues on
algorithm for the BE service the Iub interface: ATM= 4 queues,IP= 6 queues. The rest queues(except
queues with absolute priorities) use the RR( Round-Robin )scheduling
algorithm

2. If the duration for buffering the data in queue X > CONGTHDX(25ms


Default ) , and this queue enters the congestion state and the port of this
queue becomes congested accordingly. The interface boards send
congestion signals to DPU and this DPU reduce the transmission rate of
the BE service

3. When the buffer length of the queue >DROPPKTTHDX(60ms


Default ), the RNC starts discarding the subsequent packets added to
queue X from the buffer are discarded.

4. If the duration for buffering the data in queue X > CONGCLRTHD0X


(15ms Default ), the queue leaves the congestion state. The port is
recovered if all the queues on the port leave the congestion state. The
interface boards send congestion resolving signals to the associated DPU
and it restore the transmission rate of BE users on the port.

5. After the BE users leave the congestion state, the RNC increases the
transmission rate every 10 ms according to the increasing step until the
BE users reach the Maximum Bit Rate (MBR).
INT load analysis

•DSP is complete pool in the same MPU, so DSP load analysis base on MPU
Case 1 – INT load
25.00%
INT Mean CPU Load Subrack INT num
20.00% 0 0_20_0_GOUC 36
0 0_20_1_GOUC 34
15.00% 1 1_22_0_GOUC 39
1 1_22_1_GOUC 38
10.00% 1 1_18_0_AOUC 1
2 2_22_0_GOUC 40
5.00%
2 2_22_1_GOUC 41
0.00% 2 2_16_1_AOUC;2_16_3_AOUC 1
0:18 0:20 0:22 0:26 1:16 1:18 1:20 1:22 1:24 1:26 2:20 2:22 2:24 3:20 4:20 2 2_16_2_AOUC 1
SRN:SN 2 2_16_0_AOUC 1
Subrack No. Slot No. Board Type Interface Type Trans Load Current
2 2_16_1_AOUC 1
0 20 GOUc IUB 57.79% 70 2 2_16_1_AOUC 1
1 22 GOUc IUB 93.49% 77
2 20 GOUc Unused 0.00% 0 SubRack NodeB Number Cell Number
2 22 GOUc IUB 95.10% 81
2 24 GOUc Unused 0.00% 0
0 30 359
2 26 GOUc Unused 0.00% 0 1 42 486
3 22 GOUc Unused 0.00% 0 2 69 697
3 26 GOUc Unused 0.00% 0 3 52 558
4 22 GOUc Unused 0.00% 0
4 30 339
4 24 GOUc Unused 0.00% 0
5 22 GOUc Unused 0.00% 0 5 10 85
5 24 GOUc Unused 0.00% 0

 Finding: 1_22 and 2_22 with high Transload Ratio, and more than 80%. Most of the NodeB were configuration on 0_20,
1_22 and 2_22 , but other INTs were unused.
 Impact: The User plane load is close to the Max, congestion or limit the throughput will occur. Increase the load among

the subracks.
 Action: Rehoming the NodeB to according to the attachment with High PRI
Back
Flow Control – MPU overload
 MPU subsystems allocate SPU, DPU, and INT resources to Users

 EVT-22835 Flow Control is reported when the CPU usage and message block occupancy rate
reached the THD and the basic flow control function is started

Flow Control Impact on Controlling


Flow Control Object Key Action
Function Services Command

Printing flow control 1) Disable Printing, Debugging and


Logging
High CPU usage
2) Basic FC, and non-critical
or message block Debugging flow control None SET FCSW
functions
occupancy rate
3) Printing, Debugging THD= [50,60]
Logging flow control 4) Logging THD= [80,90]
1) RRC connection requests SET
MPU overload
High CPU usage 2) THD=80% (unconfigurable) Yes URRCTRLSWITC
backpressure
3) MPUSW=on (default, enabled) H
Flow Control – MPU overload
Flow Control Triggered by MPU Overload Backpressure 1. Under heavy traffic, the CPU of the MPU may be
overloaded and fail to process services properly as
a result.

2. Congestion detection is performed based on the


instantaneous CPU usage of the MPU, if the CPU

3. When the CPU usage of the MPU >=80%, the MPU


sends a congestion message to the CPUS.

4. Upon receipt congestion message, the CPUS


adjusts the flow control level( total 30 levels, A
higher flow control level means fewer RRC
connection requests admitted each second, 0 no
flow control)

5. The RNC adjusts the number of RRC connection


requests that can be admitted on the CPUS each
second according to the flow control level change.
Flow Control – MPU overload

Check whether load among MPU subsystems is


unbalanced.

 In RAN13.0 and earlier versions


• Root cause: this is because the number of boards managed
by each MPU subsystem is in great difference.
• Solution: Add new MPU subsystems or boards managed by
MPU subsystems can be re-allocated

 In RAN14.0 and later versions: MPU pool is introduced and it


is specified by the MPULOADSHARETH =50%(default)
Check whether the MPU load reaches the • Root cause: it is caused by unreasonable THD setting.
corresponding threshold: If CUP load is unbalanced, and under TH=50%, it is ok.
50 as recommended capacity expansion threshold If CUP load is unbalanced, but above TH=50%, it isn’t ok.
60 as emergency capacity expansion threshold • Solution: check the MPULOADSHARETH setting.

Solution
In RAN13.0 and earlier versions: capacity expansion
needs to be performed on each MPU subsystem
In RAN14.0 and later versions: capacity expansion is
performed for RNC level MPU subsystems
Case1– MPU overload
MPU Current MPU Banding Board Recommended MPU Banding Board load
MPU load Distribution SPUb 0:0, SPUb 0:4, SPUb 0:8, DPUe 0:14, DPUb SPUb 0:0, SPUb 0:4, SPUb 0:8, DPUe 0:14, DPUb
60 0:0:0 44.25
0:15, FG2a 0:18, FG2c 0:24, FG2c 0:25 0:15, FG2a 0:18, SPUb 0:22, FG2c 0:24
SPUb 0:2, SPUb 0:10, DPUb 0:16, DPUb 0:17,
SPUb 0:2, SPUb 0:10, DPUb 0:16, DPUb 0:17,
50 0:2:0 FG2a 0:19, SPUb 0:20, FG2c 0:25, FG2c 0:26, 43.38
FG2a 0:19, FG2c 0:26, FG2c 0:27
FG2c 0:27
40 SPUb 1:0, SPUb 1:4, SPUb 1:8, DPUe 1:25, FG2c SPUb 1:0, SPUb 1:4, SPUb 1:8, DPUe 1:25, FG2c
1:0:0 42.46
1:26, DPUe 1:27 1:26, DPUe 1:27
30 SPUb 1:2, SPUb 1:10, DPUb 1:12, DPUb 1:13, SPUb 1:2, SPUb 1:10, DPUb 1:12, DPUb 1:13,
1:2:0 DPUb 1:14, DPUb 1:15, DPUb 1:16, DPUb 1:17, DPUb 1:14, DPUb 1:15, DPUb 1:16, DPUb 1:17, 43.73
AEUa 1:18, AEUa 1:20, SPUb 1:22, AOUa 1:24 AEUa 1:18, AEUa 1:20, SPUb 1:22, AOUa 1:24
20
SPUb 2:2, SPUb 2:10, DPUb 2:17, GOUc 2:19, SPUb 2:2, SPUb 2:10, DPUb 2:17, GOUc 2:19,
2:2:0 45.52
SPUb 2:22, DPUb 2:24, DPUe 2:27 SPUb 2:22, DPUb 2:24, DPUe 2:27
10
SPUb 3:0, SPUb 3:4, SPUb 3:8, DPUb 3:12, DPUb SPUb 3:0, SPUb 3:4, SPUb 3:8, DPUb 3:12, DPUb
3:0:0 3:13, DPUe 3:14, DPUb 3:15, DPUb 3:16, AOUa 3:13, DPUe 3:14, DPUb 3:15, DPUb 3:16, AOUa 43.48
0 3:20, FG2c 3:24, GOUa 3:25, GOUc 3:26 3:20, FG2c 3:24, GOUa 3:25, GOUc 3:26
0_0 0_2 1_0 1_2 2_0 2_2 3_0 3_2 4_0 4_2 5_0 5_2 SPUb 3:2, SPUb 3:10, DPUb 3:17, DPUe 3:21, SPUb 3:2, SPUb 3:10, DPUb 3:17, DPUe 3:21,
3:2:0 46.56
SPUb 3:22, DPUe 3:27 SPUb 3:22, DPUe 3:27
VS.XPU.CPULOAD.MEAN(Busy Time) VS.XPU.CPULOAD.MAX(Busy Time)
SPUb 4:0, SPUb 4:4, SPUb 4:8, DPUb 4:12, DPUb SPUb 4:0, SPUb 4:4, SPUb 4:8, DPUb 4:12, DPUb
4:13, DPUe 4:14, DPUb 4:15, DPUb 4:16, DPUb 4:13, DPUe 4:14, DPUb 4:15, DPUb 4:16, DPUb
4:0:0 44.67
4:17, UOIa 4:22, UOIa 4:23, GOUa 4:24, GOUa 4:17, UOIa 4:22, UOIa 4:23, GOUa 4:24, GOUa
4:25, GOUc 4:26 4:25, GOUc 4:26
Average of
MPU Expansion Busy Time (As SPUb 4:2, SPUb 4:10, SPUa 4:18, DPUb 4:20, SPUb 4:2, SPUb 4:10, SPUa 4:18, DPUb 4:20,
RNCID RNCNAME
Suggestion
CPULOAD(Busy
Hour)
4:2:0 34.97
Time) AOUc 4:21, GOUc 4:27 AOUc 4:21, GOUc 4:27
SPUb 5:0, SPUb 5:4, SPUb 5:8, DPUb 5:12, DPUb SPUb 5:0, SPUb 5:4, SPUb 5:8, DPUb 5:12, DPUb
RNC_Simatupang 2015-07-29 5:13, DPUe 5:14, DPUb 5:15, AOUa 5:18, AOUa 5:13, DPUe 5:14, DPUb 5:15, AOUa 5:18, AOUa
162 N 41.98% 5:0:0 37.29
3 19:00:00 5:19, AOUc 5:22, GOUa 5:24, GOUa 5:25, GOUc 5:19, AOUc 5:22, GOUa 5:24, GOUa 5:25, GOUc
5:26 5:26

 Finding: MPU of subrack is a little high


 Suggestion: change the MPU board binding board
Flow Control – SCU overload
 SCU provides maintenance, management, and GE switching platforms for the subrack, and
implements MAC switching and interconnections between different modules

 If the link bandwidth usage >= threshold (70%) for 5 minutes, ALM-20277 Communication
Congestion Between Subracks is reported

Flow Control Impact on Controlling


Flow Control Object Key Action
Function Services Command

Printing flow control 1) Disable Printing, Debugging and


Logging
High CPU usage
2) Basic FC, and non-critical
or message block Debugging flow control None SET FCSW
functions
occupancy rate
3) Printing, Debugging THD= [50,60]
Logging flow control 4) Logging THD= [80,90]
1) THD= [75,85]
2) SCU instructs the DPU to start
High link
Inter-Subrack Flow Control flow control, reducing the Yes SET ETHSWITCH
bandwidth usage
transmit rate of BE services
gradually
Flow Control – SCU overload

Check whether SCU load is proper :


60 as recommended capacity expansion threshold

Solution
recommended that SCUa boards be replaced by SCUb
boards

E interface networking configurations be optimized to


decrease inter-subrack traffic

In most cases, SCU boards do not become a bottleneck. Therefore, SCU load analysis does not need
to be displayed when there is no load-related problem
Thank You!

You might also like