The MPLS-OPS Archive

Cell Relay Retreat>MPLS-OPS Archive>month:2004-Jan> msg00018



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

Re: Link Stagging & BDF (Bidirectional Forwarding Detection)

  • From: Eric Osborne <eosborne@cisco.com>
  • Date: Mon, 12 Jan 2004 11:35:32 -0500
  • Cc: mpls-ops@mplsrc.com
  • Resent-Date: Mon, 12 Jan 2004 12:09:01 -0500
  • To: "M. ELK" <elkou141061@hotmail.com>
  • User-Agent: Mutt/1.2.5i
  • X-GPG-Fingerprint: 6412 0836 E440 B3EA 980C 4951 611E 1819 2E71 8562

On Mon, Jan 12, 2004 at 08:01:47AM +0000, M. ELK wrote:
> 
> Hi
> 
> In the legacy FR/ATM switch world , the link initilization/stagging phase 
> test some link c/c and
> compare against some pre-configure value and/or thershold .
> as an example of C/C : Remore node ID , Link speed , quality (BER), RTD 
> ..etc .
> if the test is not passed the link will not be declared up and the link 
> remain in the stagging phase .
> 
> As another example , the analog modem with DBU (dial back up) will not drop 
> the dial line and
> revert to the recoved leased line unless the leased line pass some BER 
> Threshold .this to eliminate
> link flapping and unstability  .
> 
> In the IP world , a link flapping cause a lot of unstability on the IGP and 
> the netw even with today
> tier 1 mnafucturar box's (with knob's such as delayed SPF calculation , 
> interval between LSP generation ..etc ) which indicate that the link was 
> prematurely considered usable and
> declared up .
> 
> So my first question , what is the opinion in the list about if a link 
> stagging phase (even as optional feature) worth to be incorporated in the IP 
> world .

We have a knob called ip event dampening, sounds like it might be waht
you want.  



eric

> 
> My 2nd point , from the above we need some sort of "hello" protcol to be 
> used during the stagging
> phase . we could use the BDF (Bidirectional Forwarding detection 
> ,draft-katz-wazd-bfd-01.txt) echo
> feature , link C/C ( ex: the remote-ID (ie: remote IP address ) ,quality 
> ,rtd ) could be detected/calculated .The BFD will not declare the session up 
> unless it pass the test
> (comparing against pre-configure value and/or threshold) .
> 
> (Note : in BDF ,their is no discovery mechanism so the remote node interface 
> IP@ need to be config
>           manualy and the echo packet to be sent with TTL=2 .
>           Other possibility ,is to let the IGP run it's adjacency hello 
> protcol to aquire the remote
>            IP@ but stop short of declaring the link up . the IGP pass the 
> IP@ to the BDF to carry
>            the echo test  ,the IGP will wait the BDF declaring the seesion 
> up before using and
>            advertising the link ) .
> 
> 
> my 2nd question , is the above workable .
> 
> Note : i included Dave Katz & Dave Ward the authors of BDF as BCC ,so in 
> case they are not interested to join the discusion this step will minimize 
> the nbr's of  msg's they will recieved to just this msg.
> 
> 
> Brgds
> 
> _________________________________________________________________
> There are now three new levels of MSN Hotmail Extra Storage!  Learn more. 
> http://join.msn.com/?pgmarket=en-us&page=hotmail/es2&ST=1
> 
> -------
> The MPLS-OPS Mailing List
> Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
> Archive: http://www.mplsrc.com/mpls-ops_archive.shtml

-------
The MPLS-OPS Mailing List
Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
Archive: http://www.mplsrc.com/mpls-ops_archive.shtml