Sie sind auf Seite 1von 2

FAQ-SPU SUBSYSTEM LOAD SHARING MECHANISM

Download ( China)

Recommen Download Mirror o China d to Others o America Feedback o UK o Bahrain o Brazil

Document Properties ID: Informatio n Type: Update Time: Fault Type: Permissio n Level: Phenome non Descriptio n: Alarm Informatio n: Cause Analysis: Handling LOAD SHARING MECHANISM Process: 1when ue try to access a cell, the connection will be set up under the CPU which the cell was config ured firstly. LOAD SHARING MECHANISM will be activated only when the default CPU is overloaded. 2When the cpu is overloaded, MPU will choose the lowest loading subrack at first, then, it will choos e the lowest loading cpu in the selected subrack. The LOAD SHARING MECHANISM will be triggered under the following conditions (1) When the occupancy exceeds the cpu occupancy threshold Different kind of service has different cpu occupancy threshold, the values are as follow: NODEB sites are not configrated on subsystem averagely SE0000492992 Troubleshooting Cases Author: Quality Level: JIANG WEN B

2010-12-16 20:35:16

Views:

79

Others

Keywords:

spu cpu overload load sharing

01Huawei Engineers Permission

This is very common that some spu subsystem usage reach over 70%, while some others are less tha n 30%, how to deal with this issue? CPU overload alarm

message: RRC connection load sharing CPU control threshold is 70% message: RRC connection load sharing CPU recover threshold is 60% call:: RRC connection load sharing CPU control threshold is 85% call:: RRC connection load sharing CPU recover threshold is 75% location register:RRC connection load sharing CPU control threshold is 90% location register: RRC connection load sharing CPU recover threshold is 80% (2) When the occupancy does not exceed the cpu occupancy threshold, but reach the Forwarding thre shold of control plane load sharing.CTRLPLNSHARINGOUTTHD, and number of incoming calls in each second re aches the SharingOutCAPSMidLoad. (3) When the occupancy does not reach the Forwarding threshold of control plane load sharing, and nu mber of incoming calls in each second reaches the MAXCAPSLOWLOAD System will try to make load sharing if each one of the 3 conditions above is met. And there are also 2 preconditions for load sharing: 1)Load sharing process switch is ON 2)Callshock control switch is OFF or even callshock control switch is ON adn there is no load sharing f ailure in one second. The configuration is as below: Set the Forwarding threshold of control plane load sharing.CTRLPLNSHARINGOUTTHDand CAP S threshold: SET CTRLPLNSHAREPARA: CTRLPLNSHARINGOUTTHD=50, CTRLPLNSHARINGOUTOFFSET=5 , MAXCAPSMIDLOAD=100, SHARINGOUTCAPSMIDLOAD=10, MAXCAPSLOWLOAD=150; Set the process switch: SET RRCTRLSWITCH: PROCESSSWITCH=RNC_SHARE_SWITCH-1; Set system-level CallShock: SET UCALLSHOCKCTRL: CallShockCtrlSwitch=SYS_LEVEL-1; Suggestio Only RR message can be shared in subsystems, about 20%~25% load can not be shared, such as cell ns and resource application

Summary: instance applicationIUB interface message fowarding. thus, if there is high traffic, even load sharing can not avoid cpu overload. so the best way to slove this issue is to configrate NODEB sites on subsystem averagely.

Das könnte Ihnen auch gefallen