[cisco-nas] CPU usage : PRI vs CAS

[Pierre Nepveu]

Aaron,

once again, you save the day!  Thanks.

One more question, if I may...

Would you remind me why I would use ISDN instead of CAS if I only get
dial-up/modem calls ?  We used to use CAS and then switched to PIRs. I remember
that it's difficult to match user vs line with CAS, but it's not a whole lot
easier with PRIs. Plus, I lose one line for ISDN signaling.

So, why would I stick with ISDN ?

-------------------------------------------------------------------
Pierre Nepveu, CCNP                    tel: +1 514.380-4289 
Administrateur de reseau                    +1 888.INFOVTL x 4289
Ingenierie / Acces Internet            fax: +1 514 899-8452
Videotron Telecom Ltee (VTL) - Montreal (Quebec), Canada
-------------------------------------------------------------------


Le 2004-04-19 à 12:44, Aaron Leonard a écrit:

AL> Hi Pierre,
AL> 
AL> Let me see if I can touch on the several issues you raise here.
AL> 
AL> 1. How to keep ISDN layer 2 from dropping under load.  This was
AL> addressed by the restructuring of the ISDN code ... this went
AL> in at the end of 12.2T so 12.3 would have it ... not that that's
AL> much help for an AS5200.
AL> 
AL> As a workaround on an older system, you will want to set 
AL> "process-max-time" down low and set "scheduler interval" down low
AL> (other platforms may use "scheduler alloc".)
AL> 
AL> process-max-time 20     ! hidden in older IOS
AL> scheduler interval 200  ! use "scheduler alloc 3000 1000" on other platforms"
AL> 
AL> 2. Does CAS use less CPU than ISDN.  The answer is yes.  The
AL> issue you have though is not that ISDN is using too much CPU,
AL> it's that Nachi and friends are using too much CPU and that
AL> ISDN is fragile in the face of CPU overload.  It is true that
AL> CAS would be less fragile.  On the CAS vs. PRI modem issue ...
AL> V.90 definitely works over CAS; all other things being equal,
AL> calls will typically train at a 1333bps lower rate when using
AL> CAS vs. PRI.
AL> 
AL> 3. (Not asked but implicit).  How best to mitigate the impact
AL> of infected clients on the AS5200.  People with field experience
AL> will be able to provide some useful hints I hope; I can point
AL> you to the "Nachi Worm Mitigation Recommendations" page at
AL> http://www.cisco.com/warp/public/707/cisco-sn-20030820-nachi.shtml .
AL> 
AL> Aaron
AL> 
AL> ---
AL> 
AL> > anybody has any thought on how to reduce CPU usage on AS5200's ?  The problem
AL> > arises when a client infected with worms such as Nachi gets online. They end up
AL> > crashing the NAS because they generate so many small packets that keep the cpu
AL> > busy. The NAS cannot keep it's connection to the telco switch (or the isdn
AL> > process crashes, or whatever). Anyway, the NAS becomes useless and need to be
AL> > rebooted.
AL> 
AL> > Any thoughts on how to circumvent the problem ?
AL> 
AL> > Would using CAS instead of PRI/ISDN aleviate the problem (the ISDN process
AL> > wouldn't crash, for one thing) ?
AL> 
AL> > Would using CAS make V.90 unusable (if I rebember correctly, K56 was ok) ?
AL> 
AL> > Thanks !
AL> 
AL> > -------------------------------------------------------------------
AL> > Pierre Nepveu, CCNP                    tel: +1 514.380-4289
AL> > Administrateur de reseau                    +1 888.INFOVTL x 4289
AL> > Ingenierie / Acces Internet            fax: +1 514 899-8452
AL> > Videotron Telecom Ltee (VTL) - Montreal (Quebec), Canada
AL> > -------------------------------------------------------------------
AL> 
AL> 
AL> > _______________________________________________
AL> > cisco-nas mailing list
AL> > cisco-nas at puck.nether.net
AL> > https://puck.nether.net/mailman/listinfo/cisco-nas
AL>