Cisco 7206 IOS for PPPoE Termination

Hello everybody,
I am using C7206 VXR NPE-G2 routers as BRAS in my network and the current
IOS is *c7200p-adventerprisek9-mz.124-24.T.bin* on them.
Also their memory upgraded to 2GB instead of 1GB.
And I have near 6500 online user on each of my BRAS and there is no
speciefic feature except aaa with radius and ordinary features.
There router is also terminating dot1q too because my PSTN centers traffic
comes through dot1q vlans to BRAS es.
I think I have some problem with current IOS, My CPU Usage is abnormal and
Its near %70 or %80.
And when I have a network problem and some of PSTN centers goes down CPU go
to %99 and it gets problem to recovery.
Do you know any good IOS for me as a service provider to use?
I heard that some service providers have near 8000 online user on 7206.

6000 PPP users on a NPE-G2 is way too much imho. Currently we do no more than 3000 users on a NPE-G2 with PPPoA. (Max cpu 50%).
5 years ago, we did about 5000 users on a NPE-G2, but as traffic ratio's grow each year the maximum users a NPE-G2 can handle will drop each year. Don't forget an NPE-G2 is a software based plaform, so traffic forwarding is done in software CPU.

Rinse Kloek
Op 23-9-2012 20:51, Shahab Vahabzadeh schreef:

Which software you used before for them?

For this application, you may wish to consider the service provider images.

The latest 15.x(S) image works, as it is the derivative of what was
formerly the service-provider oriented 12.2(SRx) images.

However, it's unlikely to drop steady state CPU, but it may contain some
optimizations for concurrent PPP (re)negotiations on the G2 platform during
session recovery.

PPPoE will generally handle more users on ethernet as it is easier to push
packets on when not dealing with the ATM encapsulations, but to what extent
this holds true on the 7200, I can't tell you for sure.

I'd also read the broadband aggregation guide under the IOS documentation
on, and tune all the knobs that may help you, there are some
pointers on what items on virtual-templates are punitive in performance,
other optional items such as disabling SNMP counters on virtual access
interfaces to reduce cpu usage, and other items that may help little by
little. There are also various knobs to throttle PPPoE renegotiation rates
during recovery.

I wish you luck (and consider getting another and/or bigger router to split
the load).

Dear Paul,
Thanks for you reply, May I have those optimization knobs for
virtual-template and throttles?
Maybe looking into your configurations help me in this field.
I will look for the service provider image too.

You are joking I hope

why joking Mark?

I know how to use Google dear Mark, but I mean which configuration is
working succesfully in their network.
I am currently using this config:

bba-group pppoe TEST
virtual-template 1
sessions per-mac limit 2
sessions per-vlan limit 5000
sessions per-vc throttle 15 30 300
sessions per-mac throttle 15 30 300
sessions auto cleanup