Re: [Int-area] new I-D available - tunnel issues

[Alexandru Petrescu <[e-mail]>]

Thanks for this draft.

FRom a mobility (mip-based) standpoint.

-network mobility (the mip-based nemo type) may easily involve several
  levelsof encapsulation, think networks moving into networks, this
  exacerbating the fragmentation and mtu effects and pmtu necessity.
-'stalemate' situations occured in a lab with two moving networks,
  described in appendix B of rfc4888, due to the impossibility of setting
  up the mip tunnels, because of the way things moved.  Never noticed
  that in a non-lab deployment though.
-it's good to note when tunnels are _not_ as bad as feared.  For
  example, contrary to some belief, a single encapsulation to HA
  instead of straight communication does not slow down the application,
  nor does it consume excessive bandwidth nor does it slow the handovers.
-qos: there's currently no means to perpetrate the qos decision of a
  flow (tc/flow label) beyond the HA.  This is potentially an issue with
  end-to-end architectures: a mobile end would like to maintain the tc/fl
  decisions it makes  beyond the HA (the one encapsulating/decapsulating
  its traffic when away from home), and not let the HA make this
  decision.

Virtual interfaces implementing tunnels often have a problem with
implementing link-layer multicast.  They're interfaces, but they're
virtual.  What they lack from real ones is link-layer multicast.  Given
how much IPv6 relies on link-layer multicast often protocols like ND,
DHCPv6, ICMPv6 and MIP6 are quirky implementations over tunnels.

Beyond that, there's a very much fundamental problem with tunnels, I
mean the shift and leap of semantics, overlays.  I'm sure some of this
is at the origin of this draft, but I'm not sure whether the immediate
effects of this problem are listed.  An immediate effect is applicable
to MIP as well as to the popular VPN usage: should an end's default
route point to the tunnel or to the naked first-hop router?

Just some thoughts...

Alex

Joe Touch wrote:
> Hi, all,
> 
> I and Mark Townsley recently completed an I-D summarizing our 
> presentation from the last IETF on tunnel issues.
> 
> Comments welcome, of course. FYI, a related document on the IPv4 ID 
> will be noted to this list shortly.
> 
> Joe and Mark
> 
> ------------------------------------- Filename: 
> draft-touch-intarea-tunnels Revision:     00 Title:         Tunnels 
> in the Internet Architecture Creation_date:     2008-07-07 WG ID: 
> Independent Submission Number_of_pages: 21 
> http://www.ietf.org/internet-drafts/draft-touch-intarea-tunnels-00.txt
> 
> 
> 
> 
> Abstract: This document discusses the role of tunnels in the Internet
>  architecture. It explains their relationship to existing protocol 
> layers, and the challenges in supporting tunneling.
> 
> 
> ------------------------------------------------------------------------
> 
> 
> 
> 
> _______________________________________________ Int-area mailing list
>  [e-mail] https://www.ietf.org/mailman/listinfo/int-area


______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email 
______________________________________________________________________
_______________________________________________
Int-area mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/int-area