Re: Re: RR checks to avoid DoS attacks
"James Kempf" <[email protected]>
| Newsgroups | gmane.ietf.mobike |
|---|---|
| Message-ID | <[email protected]> |
Tero,
If the issue is whether the gatway can continue sending packets to a known
valid address while it is testing validity of a new one, I have no
objection. I think this is fairly straightfoward and logical, and I can't
see any downside. It is just a matter of determining when the state machine
allows the gatway to cut over to the new address. Naturally, there could be
packet loss on the old address if the host moves out of coverage range
(like, for example, the reverse transition when a host moves from WLAN to
GPRS because the hotspot coverage area is running out), but that shouldn't
be the concern of MOBIKE, that's up to the access network to handle.
jak
----- Original Message -----
From: "Tero Kivinen" <[email protected]>
To: "James Kempf" <[email protected]>
Cc: <[email protected]>; <[email protected]>;
<[email protected]>; <[email protected]>
Sent: Friday, May 07, 2004 5:09 AM
Subject: Re: [Mobike] Re: RR checks to avoid DoS attacks
> James Kempf writes:
> > This sounds to me more like something the mobility mangement protocol
should
> > be doing. There really is no way, in any case, to avoid having packets
in
> > the pipeline between the gateway and the host when the change occurs if
the
> > gateway is dealing with streaming media like video. These packets will
be
> > lost unless the access routers in the mobile host's access network take
some
> > kinds of measures to ensure that the packets are properly forwarded to
the
> > new address.
>
> I was more referencing the case where host can still receive from both
> interfaces, i.e. it is switching from the GPRS link to the WLAN link,
> but can still receive packets in the GPRS link after the transtion. It
> would be annoying if the gateway would stop sending me packets while
> it is doing the RR for my WLAN address, while it could still sent them
> to my GPRS address, and get them through.
>
> So I would propose that we could have option saying that if the
> address_update list still includes the currently used address, but the
> preferred address changes, then the currently used address is still
> used to send traffic until the RR of the new preferred address is
> done, and the new preferred address is used after the RR is
> successful.
> --
> [email protected]
>