[c-nsp] PCF-PDSN

[Vikas Sharma]

Hi,

Can u let me know what algorithm Lucent is using for this purpose? What I
found from my system it is using ping to check the PDSN availability.
Suppose if link flaps and there is a packet drop of 5 packets, PCF will
choose the next available PDSN. But in reality first PDSN is not down, there
was just a link flap. That's how it generate overlapping billing records.

It will be really good if u can share with me the logic/ algorithm used by
Lucent.

Thanks
Vikas Sharma



On 4/11/06, Vikas Sharma <vikassharmas at gmail.com> wrote:
>
>  Hi,
>
> This feature I have seen in Cisco PDSN. They support geographical
> redundancy by implementing controllers (active / standby in HRRP mode). All
> PCF has ip of this controller and they connect to this first. Tha controller
> decide which PDSN ip should be sent back to the PCF. Again PCF does A10/A11
> request and connect to required nearest PDSN (ip provided by active
> controller). This result in overcoming inter PDSN handoff and provide
> geographical redundancy.
>
> Regards
>  Vikas Sharma
>
>
>  On 4/11/06, Alexander Serkin <als at cell.ru> wrote:
> >
> > Vikas Sharma wrote:
> > > Hi Alex,
> > >
> > > It has been observed that the moment we define primary and Secondary
> > > PDSN at PCF , it results in overlapping billing records for the
> > > subscriber.This is because the algorithms / Logic implemented at PCF
> > for
> > > declaring the Primary PDSN as OSS and Redirecting the subsequent
> > > connections to the Secondary PDSN is not full proof.
> > > This results in data session on both the PDSN simultaneously and there
> > > by Overlapping billing records.
> >
> > Well, i think this is specific to Ericsson or Motorola or both.
> > We use Lucent 5ESS with PHE2-PCF cards in it. Every PDSN is registered
> > on every PCF and PCFs are using pseudo-random PDSN selection algorithm.
> > So when one of the PDSNs come out of service all PCFs stop registering
> > sessions on it and continue using the others. Live sessions never hadoff
> > between PDSNs. That's why overlapping does not occour.
> > I never heard about the redundancy feature You describe. Is it mentioned
> > somewhere in 3GPP2 specs?
> >
> > >
> > > Pls let me know if u need more info.
> > >
> > >
> > > Regards
> > > Vikas
> > >
> > > Date: Mon, 10 Apr 2006 16:46:26 +0400
> > > From: Alexander Serkin < als at cell.ru <mailto:als at cell.ru>>
> > > Subject: Re: [c-nsp] PCF-PDSN
> > > To: cisco-nsp at puck.nether.net <mailto: cisco-nsp at puck.nether.net>
> > > Message-ID: < 443A53A2.1040104 at cell.ru <mailto:
> > 443A53A2.1040104 at cell.ru>>
> > > Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> > >
> > > Hi, Vikas.
> > >
> > > Vikas Sharma wrote:
> > >>  Hi Ronala,
> > >>
> > >>  No, Ericsson and Motorola. Right now I am not concentrating on PDSN
> > >>  clustering. Through PCF I want to achieve geographical redundancy.
> > >>  Cisco PDSN support this feature of geographical redundancy.
> > >
> > > If You have several PDSNs registered at PCF then when one of them
> > comes
> > > out of service PCF should take the next one.
> > > And new packet data session should start on the next PDSN.
> > > What do you mean by overlapping billing records?
> > >
> > >>
> > >>  Regards
> > >>  Vikas Sharma
> > >>
> >
> >
> > --
> > Sincerely Yours,
> > Alexander Serkin,
> > Moscow Cellular Communications,
> > ph. +7(495)7952089
> > fa. +7(495)7952084
> >
>
>