SCTP Discussions

Text archives Help


[SCTP-Discussion] Re: Another situation on Notify Message Treatment


Chronological Thread 
  • From: "Brian F. G. Bidulock" <bidulock AT openss7.org>
  • To: Sundaresan M <sundarem AT future.futsoft.com>
  • Cc: discussion AT sctp.de
  • Subject: [SCTP-Discussion] Re: Another situation on Notify Message Treatment
  • Date: Mon Dec 31 07:22:01 2001
  • Dsn-notification-to: <bidulock AT openss7.org>
  • List-id: SCTP Discussions <discussion.sctp.de>
  • Organization: http://www.openss7.org/

Sundaresan,

On Mon, 31 Dec 2001, Sundaresan M wrote:

> Hi Brian and group,
>
> Thanks for the reply. Now I have an entirely different situation.
>
> I am working with Version 06 of M3UA. In this version we dont have the
> concept of ASP Identifier to identify individual ASP's. So how can this be
> managed (I am designing a conformance suite for M3UA-06 and hence I cannot
> use latest versions of M3UA, I have to stick to version 06 only).

You would think that 3GPP would have been intelligent enough to not
reference an IETF draft. Nevertheless, for any version of M3UA draft,
ASP Identifier is not required in an ASP Up in situations where the the
SGP can always identify the ASP using exotropic means such as the IP
address and SCTP port number of the connecting ASP.

>
> As u have said I too was designing the Notify message so that it is sent to
> every ASP in the AS except for those ones in the Down state. But, imagine
> this situation.
>
> Configuration: Two ASP's and One SG.
> ASP1: Already in Active State
> ASP2: Working in Override mode, Comes to Active State.
>
> Now once ASP2 comes to Active state SG will send Notify to both the ASP's
> with an Status Identification "Alternate ASP Active".
>
> I understand that now SG will stop sending the traffic to ASP 1 from now on.
>
> My questions are: If there is no ASP Identifiers existing to identify ASP's
> then how will the ASP1
> 1) differentiate that he has to stop processing traffic from that SG and
> will take further steps to either bring itself to Inactive state if is not
> serving anyother AS.
> 2)know that it is only ASP2 which took over the traffic. But before this, is
> there a need for ASP1 to know that ASP2 has taken over the traffic for that
> SG.

Well, that was why ASP Identifier was added to M3UA. If you wish to
used the broken M3UA-06 (and--believe me--it is broken: just see the
comments on the SIGTRAN mailing list for M3UA-06 to see in just how many
ways it is broken...) then ASP1 will have to determine from ASP2 using
an unspecified (and therefore, untestable) ASP to ASP communication
protocol.

>
> 3) Since ASP2 also recieves this Notify message what if he thinks that some
> other ASP had taken over the traffic.

Even in the broken M3UA-06, ASP2 should know that it has become active
because it sent the ASPAC that cause the override. If you are thinking
that ASP2 might have a race with, say, ASP3, then it falls back to the
unspecified ASP-to-ASP communication ensure that things are done
atomically and that there are no races.

In M3UA-10 (extremely close to RFC) even if multiple ASPs race, or get
partitioned from each other so that ASP-to-ASP communication is not
possible, each ASP is notified of which ASP has taken over in the
override group via the ASP Identifier in the NTFY.

If you have any choice whatsoever, I would spend my time lobbying 3GPP
to update to the RFC version of M3UA and remove references to expired
internet drafts.

If you are developing test cases for M3UA-06, you might want to take a
look at the M3UA-05 test spec that is still available on the IETF server
at:

ftp://ftp.ietf.org/internet-drafts/draft-hss-test-spec-m3ua-00.txt

The author of this draft (Ashok Singh at HSS) might also have a M3UA
draft 6 version of the document, but you will have to contact Hughes for
that. The email address is in the draft.

Hope that helps.

--brian

>
> Best Regards,
> Sundar.
>
>
>
> -----Original Message-----
> From: Brian F. G. Bidulock
> [mailto:bidulock AT openss7.org]
> Sent: Sunday, 30 December 2001 6:38 PM
> To: Sundaresan M
> Cc:
> discussion AT sctp.de
> Subject: Re: [SCTP-Discussion] Notify Message Treatment
>
>
> Sundaresan,
>
> On Sat, 29 Dec 2001, Sundaresan M wrote:
>
> > Hi,
> >
> > I would like to know how the Motify messages are treated in the Adaptation
> > layers like IUA, M3UA etc.
> >
> > QUESTION 1
> > **********
> > Configuration 1: One ASP and One SG
> > -------------------------------------
> > ASP is moving from Inactive state to Active state. Now the AS state
> changes
> > from AS_INACTIVE to AS_ACTIVE. Will a Notify message be sent by the SG to
> > this one and only ASP in the Configuration. Will this treatment change
> > depending on the AS_STATE change also. What I mean is, if the state
> changes
> > from Active to Inactive will this ASP(being the last one in the AS)
> receive
> > a Notify(AS_PENDING)!!
>
> "A Notify message reflecting a change in the AS state MUST be sent
> to all ASPs in the AS, except those in the ASP-DOWN state, with
> appopriate Status Information and any ASP Identifier of the failed
> ASP." - M3UA-10
>
> A Notify("AS-Pending") "MUST" be sent to the ASP.
>
> >
> > QUESTION 2
> > ***********
> > Configuration 2: Two ASP's (ASP1 and ASP2) with One SG:
> > ------------------------------------------------------
> > ASP 1 : Active
> > ASP 2 : Inactive
> >
> > Now ASP1 moves from Active to Inactive due to LM primitives. Will the
> > Notify(AS_PENDING) message be sent to Both ASP1 and ASP2 serving the SG?
> or
> > will it be sent only to ASP2.
>
> "If no other ASPs in the Application Server are in the state
> ASP-ACTIVE, the SGP MUST send a Notify message ("AS-Pending") to
> all of the ASPs in the AS which are in the state AS-INACTIVE."
> - M3UA-10
>
> The message "MUST" be sent to both ASP1 and ASP2.
>
> >
> > QUESTION 3
> > **********
> > Will the AS_STATE be maintained in the database by the individual ASP's.
>
> The ASP will need to know what state it is in for each AS that it serves.
>
> >
> > QUESTION 4
> > **********
> > Some Stack implementors say that NOTIFY message treatment is optional.
> Does
> > this mean that the usage of the Notify itself is optional. If it is so
> then
> > how will the ASP's know the AS_state and will act on AS_State changes?
>
> Sending of the Notify message is not optional. The response of the ASP to
> the
> Notify message is not proscribed. There is a difference. Notify messages
> are considered advisory in that it is up to the ASP to consider the
> information in the Notify message and other information that it has at its
> disposal (such as a knowledge of the state of other ASPs) before deciding
> what
> action to take.
>
> Hope that helps.
>
> --brian
>

--
Brian F. G. Bidulock ¦ The reasonable man adapts himself to the ¦
bidulock AT openss7.org
¦ world; the unreasonable one persists in ¦
http://www.openss7.org/ ¦ trying to adapt the world to himself. ¦
¦ Therefore all progress depends on the ¦
¦ unreasonable man. -- George Bernard Shaw ¦




Archive powered by MHonArc 2.6.19+.

Top of page