[c-nsp] High Availability for CGSE - NAT44 - Inter Chassis Redundancy

Pshem Kowalczyk pshem.k at gmail.com
Wed Jun 1 02:43:03 EDT 2011


Hi,

On 31 May 2011 08:33, cisco-pe <cisco.pe.1977 at gmail.com> wrote:
> Hello Guys,
> We are planning to deploy NAT44 in a pair of CRS-1/16 (Toll gates) located
> on differentes sites, using CGSE card.
> Could you help us with some recommendations about how to implement the NAT44
> high availabilty using CGSE modules on CRS-1 ?

I'm unable to decipher your diagram, but with NAT you have to take the
following things into consideration:
1. Currently CGSE does not offer hot-standby so, even if you have two
in one chassis the sessions will have to be re-established (take the
number of sessions into consideration if you do not want very long
outage)
2. With inter-chassis fail-over you have to relay on routing protocols
convergence and ability to trigger routing convergence if only the
CGSE card fails, but not the RP or uplink line cards.
3. Do you want the CGSE to participate in MPLS or not, if so are you
running TE or only LDP?
4. When the sessions are dropped some customers' sessions will not
come back automatically, user intervention will be necessary.
Depending on your customer base this might not be a big problem.
5. How are you going to scale the solution when number of
subscribers/sessions increases?
6. Do you have to record all mappings between inside/outside IPs (for
example for lawful intercept, charging or some other reasons)?

IMHO the convergence time will greatly depend on the protocols you're
natting and number of flows.

kind regards
Pshem


More information about the cisco-nsp mailing list