Are these limits based on HW performance, or are they just phase one 'safe'
levels which could be pushed up with experience?
>are inside tunnels. A tunnel is formed between LACs and LNSs. It is
>possible to have empty tunnels, since tunnels are not torn down when all
>sessions within a tunnel are terminated.
I understand that (we're using L2TP in our dial network now) - how/when does
a PM handle tearing down the tunnel itself. Tunnels can be established on
demand, and it wouldn't be good if a PM became filled with tunnels which
were idle - of course. Perhaps closing the oldest idle tunnel when a new
tunnel is requested and the limit is reached. *shrug*
>The limits are hard limits for the max number of sessions and tunnels. The
>LNS terminates L2TP with virtual ports which is a fixed number.
Now I'm confused/curious - is there a limit on the number of IPs in the pool?
What if I am using a PM-3 to handle 2 PRIs and I want to use it as an LNS.
Could I have say 128 IPs to cover both the physical ports and the LNS
virtual ports? Or is there a SW limit on mixed use? (Dare I toss IPSec
in here? How about an IPSec session termination on a PM-3 via an L2TP
tunnel? ;-) )
>>>The PM4 can handle 500 Sessions per LNS card.
>This is true the LNS card, which is just like a Quad-T1 (just without line
>ports and without modems), will terminate upto 500 sessions.
Ah, this isn't what I was thinking. I was thinking of the discussed
mega-CPU farm special card that had been discussed. Thanks for clearing that
up.
>The Quad-T1 will support upto 94 L2TP sessions.
This is a hardcoded SW limitation then? If the LNS card is the same as
the QT1, sans interfaces, then running a QT1 eith no lines would be about
the same...
Same question as for the PM-3 - are these 94 sessions in conjunction with
the up to 96 'physical' ports.
>NOTE: The PM4 as a whole can support upto 100 tunnels. So the session
>limit is card by card, while the tunnel limit is by chassis.
If the tunnels are by chassis, but sessions by card, is it safe to presume
the sessions will be dynamically assigned to an available LNS/QT1 card as
needed?
>>I'd probably use 8 with 2 NMCs just to be safe - redundancy.
>The backup SMM feature will be available in a future release, but not in
>the next open beta of the PM4.
Cold failover is still scheduled for ComOS 4.1 though, right?
-MZ
-- -=*X I'm going down... under that is! <URL:http://www.aussie-isp.net/> X*=- <URL:mailto:megazone@megazone.org> Gweep, Discordian, Author, Engineer, me.. Join ISP/C Internet Service Providers' Consortium <URL:http://www.ispc.org/> "A little nonsense now and then, is relished by the wisest men" 781-788-0130 <URL:http://www.megazone.org/> <URL:http://www.gweep.net/> Hail Discordia! - To unsubscribe, email 'majordomo@livingston.com' with 'unsubscribe portmaster-users' in the body of the message. Searchable list archive: <URL:http://www.livingston.com/Tech/archive/>