RNC RRC Load Control

RNC RRC Load Control (RRLC) supports an early rejecting filter of RRC

Connection requests in the RNC. It detects the RRC admission load peak

build up in an MC and cuts the peak growth to bring the MP load level back

to a sustainable level. The RRLC early (before causing processor overload)

rejects a fraction of the RRC connection requests, but let through the rest of the

requests for further processing. Emergency calls and CS calls are not rejected

by RRC Load Control filter. The early filter can be enabled for each cell.

Feature Designation indicates which instance of the RncFeature MO

is used to administratively control the feature. For the RRLC it is

RncFeature=RrcLoadControl. To enable the feature the attribute

rrcLcEnabled under UtranCell is used.

Description

The RRLC consists of a filtering and monitoring phase. A fraction of the RRC

connection requests that are early rejected is defined by Early Filter Rejection

Ratio. RRC Rejection Ratio (RRR) is used as an input for calculating Early

Filter Rejection Ratio. Early Filter Rejection Ratio is based on measurement

of RRR done during the previous time interval. RRC Connection requests are

semi-randomly rejected during the current time interval at the start of the time

interval. Percentage of rejected requests is defined by Early Filter Rejection

Ratio and is not changing during the interval.

Observability

The counter pmNoRejRrcConnRrcLc under UtranCell MO shows total number

of RRC Connection attempts rejected early in the RNC due to RRLC filter.

For recording the total time during which RRLC filter is triggered for each cell

pmTotalTimeRejRrcConnRrcLc can be used.

MP Load Control

The purpose of the Main Processor (MP) load control function is to protect its

MPs from overload. Good overall overload protection for the entire node is

reached by making sure that the MPs are not overloaded. MP load control

works within the context of a single MP and is contained in a separate load

module. Traffic management software executing on the same MP negotiates

with load control during the initial phases of costly operations, requesting

processor capacity as needed. Each MC is protected by its own instance of

load control program.

For the General Performance Event Handling (GPEH) load, a specific

protection feature exists that ensures that the load incurred by GPEH does not

affect traffic on the MCs – before allowing this to happen, GPEH events are

discarded. This protection mechanism is active both on the C1 MPs and on

all MCs. If GPEH events are to be discarded, an internal fault is written to the

GPEH ROP file. GPEH events are deactivated when the following load limits of

the processors are met:

EPB1

Low priority GPEH event 89%

High priority GPEH event 95%

The RNC load control mechanisms display the basic behaviour seen in Figure

3. The node has a particular capacity. When the offered traffic corresponds to this

capacity, the maximum throughput is achieved.