Cisco Cisco ASR 5000 Guía Para Resolver Problemas

configured rate. When the queue overflows (due to high incoming rate), based on the "action" configured,
packets are either dropped or rejected.

queue−size
Defines the maximum size of the pacing queue used for buffering the packets. Must be an integer from 250 to
25000. The default is 2500.

Sample Configuration

network−overload−protection mme−new−connections−per−second 400 action attach

reject−with−emm−cause no−suitable−cell−in−tracking−area tau

reject−with−emm−cause no−suitable−cell−in−tracking−area fwd−reloc drop

Now the call rate per second is set to 400 and the action is intelligent reject with cause #15 to make the User
Equipment (UE) reconnect to different Radio Access Technologies (RATs). The wait time is set to the default
(5 seconds) and the queue size is 2500.

Note: The action "reject" with EMM cause #15 "no−suitable−cell−in−tracking−area" is preferred as the calls
rejected with #15 mostly will not rearrive to MME and will go to different RAT layers (3G, 2G). The action
"Drop" for Serving Radio Network Subsystem (SRNS) relocation is for future use and will prevent a quick
reattach to MME after rejection.

Network Overload Protection: Paging Throttling

Paging Throttling protects internal MME resources (mmemgr) as eNodeB/radio resources (if needed). This
rate limit threshold shall be applicable to all the eNodeB that associates with MME for a given ASR 5000
chassis. S1 Paging requests to an eNodeB shall be rate limited at this threshold value. S1 Paging requests to an
eNodeB that exceed this threshold shall be dropped.

For MME the throttling rate calculation is relatively simple, take the standard egress paging rate in the
network plus the tolerance. (This is based purely on the design team's decision.)

Example

In busy hours, each MME handles up to 35000 paging messages per second. In case of one site failure, up to
70000 pages per second might go from one MME. Under this paging rate, the MME utilization (mmemgr)
rises close to 80% and 70000 to 80000 pages per second would be an optimal level to limit the throttling rate
in order to avoid excessive S1 signaling over mmemgr.

However, the rate is limited per average eNodeB. The average rate per eNodeB (in case of 6500 eNodeB) is
10 pages per second. However, the Tracking Areas (TAs) are not equal in the number of subscribers and
various TA/member eNodeB are loaded with paging differently. In the case of two times the difference in TA
size versus the average number of subscribers per TA, the rate per eNodeB would be 20. In the case of 20
times the difference in the TA size versus the average number of subscribers per TA, the rate per eNodeB
would be 200. This means that the feature becomes most efficient in cases when TA (in number of
subscribers) are evenly loaded.

Another action which should be taken in parallel is to activate the Intelligent Paging. Refer to the "TAI mgmt
db and LTE Paging" section in the ASR 5000 MME Administration Guide.

The configuration command is as follows:

asr5000(config)# network−overload−protection mme−tx−msg−rate−control enb s1−paging

<rate in messages per second>