Thread: [mpls-linux-general] Re: mpls-linux-general digest, Vol 1 #685 - 1 msg
Status: Beta
Brought to you by:
jleu
Menu
▾
▴
mpls-linux-general
|
From: Rakesh <ra...@ii...> - 2004-09-23 12:16:54
|
mpls may drop packets but it depends on how you configure it. If there is
any que mechanism and other LSP is configured then it can follow that new
path.
Regards,
---------------------------------------------------------------------------
| ra...@ii...
Rakesh Deshmukh | R - 314 H -3
3rd Year BTech | IIT Powai
IIT Bombay | Mumbai - 76
---------------------------------------------------------------------------
> Send mpls-linux-general mailing list submissions to
> mpl...@li...
>
> To subscribe or unsubscribe via the World Wide Web, visit
> https://lists.sourceforge.net/lists/listinfo/mpls-linux-general
> or, via email, send a message with subject or body 'help' to
> mpl...@li...
>
> You can reach the person managing the list at
> mpl...@li...
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of mpls-linux-general digest..."
>
>
> Today's Topics:
>
> 1. Increase MPLS link capacity ? (pa...@vs...)
>
> --__--__--
>
> Message: 1
> Date: Wed, 22 Sep 2004 09:50:12 +0200
> From: pa...@vs...
> To: mpl...@li...
> Cc: its...@ya...
> Subject: [mpls-linux-general] Increase MPLS link capacity ?
>
> Hello Leu and others,
>
> I have the following doubt.
>
> in an MPLS cloud / network , in an LSP , if the link
> /path capacity is insufficient, how would it be met?
>
> i.e., if the demand increases and the bandwidth in the
> LSP is insufficient how would this scenario be met ?
>
> -switch to another LSP ?
> -have a backup LSP ?
>
> comments on this ?
>
> Pathma
>
>
>
>
>
>
> --__--__--
>
> _______________________________________________
> mpls-linux-general mailing list
> mpl...@li...
> https://lists.sourceforge.net/lists/listinfo/mpls-linux-general
>
>
> End of mpls-linux-general Digest
>
|
|
From: Stefano A. <sta...@un...> - 2004-09-24 16:17:12
|
Hi all, we are interested in multiple FTN-NHLFE associations. Our goal is to split a single flow (identified by means of source and dest address and port, protocol) over different LSPs. According to RFC 3031, the FTN may map a particular label to a set of NHLFEs, but exactly one element of the set must be chosen before the packet is forwarded. Does mpls-linux already contain some code about such functionality? If not, any hint on how to proceed? Thanks a lot in advance, Stefano -- Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/ |
|
From: James R. L. <jl...@mi...> - 2004-09-24 16:57:49
|
No this functionality does not exist but I have wanted it for sometime. I already have ideas in my head of how I would like to do it. I'll try not to impose my views too much but here is where you should start. First, make sure to get the latest version of my code. I plan on releasing it soon, but I haven't finished the release process yet. You can grab the latest from my P4 tree (if you nee help doing this let me know I can send you an email with all of the details). Second I envision this feature being an instruction which could be attached to a NHLFE or an ILM. Currently there is a 'FWD' instruction. Which fowards to just one NHLFE. I think a "MFWD" or something like that, would potentially forward to mutiple NHLFE. Flags that are part of this instructions would dictate if the MFWD required replications (multicast case) or an algorithm to determine how to choose from the list of NHLFE. If your interested in different algorithms to try let me know I have a couple in mind. I can give you more specific pointers into the code if you would like. On Fri, Sep 24, 2004 at 06:16:56PM +0200, Stefano Avallone wrote: > Hi all, > > we are interested in multiple FTN-NHLFE associations. Our goal is to split > a single flow (identified by means of source and dest address and port, > protocol) over different LSPs. According to RFC 3031, the FTN may map a > particular label to a set of NHLFEs, but exactly one element of the set > must be chosen before the packet is forwarded. Does mpls-linux already > contain some code about such functionality? If not, any hint on how to > proceed? > > Thanks a lot in advance, > Stefano > > -- > Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/ > > > ------------------------------------------------------- > This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 > Project Admins to receive an Apple iPod Mini FREE for your judgement on > who ports your project to Linux PPC the best. Sponsored by IBM. > Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php > _______________________________________________ > mpls-linux-general mailing list > mpl...@li... > https://lists.sourceforge.net/lists/listinfo/mpls-linux-general -- James R. Leu jl...@mi... |
|
From: James R. L. <jl...@mi...> - 2004-09-24 16:59:27
|
I spoke too soon, someone is working on it. I will make sure to post to the list when I have patches for it. On Fri, Sep 24, 2004 at 06:16:56PM +0200, Stefano Avallone wrote: > Hi all, > > we are interested in multiple FTN-NHLFE associations. Our goal is to split > a single flow (identified by means of source and dest address and port, > protocol) over different LSPs. According to RFC 3031, the FTN may map a > particular label to a set of NHLFEs, but exactly one element of the set > must be chosen before the packet is forwarded. Does mpls-linux already > contain some code about such functionality? If not, any hint on how to > proceed? > > Thanks a lot in advance, > Stefano > > -- > Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/ > > > ------------------------------------------------------- > This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 > Project Admins to receive an Apple iPod Mini FREE for your judgement on > who ports your project to Linux PPC the best. Sponsored by IBM. > Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php > _______________________________________________ > mpls-linux-general mailing list > mpl...@li... > https://lists.sourceforge.net/lists/listinfo/mpls-linux-general -- James R. Leu jl...@mi... |
|
From: Stefano A. <sta...@un...> - 2004-09-27 15:44:37
|
Hi James and all, we agree with your view, we were thinking of an instruction for the NHLFE table. May we have more information about the current work? Maybe we can contact the authors to exchange views/contribute to the work,... TIA, Stefano Scrive "James R. Leu" <jl...@mi...>: > I spoke too soon, someone is working on it. I will make sure to > post to the list when I have patches for it. > > On Fri, Sep 24, 2004 at 06:16:56PM +0200, Stefano Avallone wrote: > > Hi all, > > > > we are interested in multiple FTN-NHLFE associations. Our goal is to split > > > a single flow (identified by means of source and dest address and port, > > protocol) over different LSPs. According to RFC 3031, the FTN may map a > > particular label to a set of NHLFEs, but exactly one element of the set > > must be chosen before the packet is forwarded. Does mpls-linux already > > contain some code about such functionality? If not, any hint on how to > > proceed? > > > > Thanks a lot in advance, > > Stefano > > > > -- > > Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/ > > > > > > ------------------------------------------------------- > > This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 > > Project Admins to receive an Apple iPod Mini FREE for your judgement on > > who ports your project to Linux PPC the best. Sponsored by IBM. > > Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php > > _______________________________________________ > > mpls-linux-general mailing list > > mpl...@li... > > https://lists.sourceforge.net/lists/listinfo/mpls-linux-general > > -- > James R. Leu > jl...@mi... > ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. |
|
From: James R. L. <jl...@mi...> - 2004-09-27 17:07:38
|
On Mon, Sep 27, 2004 at 05:48:28PM +0200, Stefano Avallone wrote: > Hi James and all, > > we agree with your view, we were thinking of an instruction for the NHLFE table. > > May we have more information about the current work? Maybe we can contact the > authors to exchange views/contribute to the work,... Christophe Fillot <cf...@ut...> is working on the code and has written some operational code in kind of a 'proof of concept' form. He and I discussed what the possible next steps are. Maybe your input could help influence this process. What are you looking for, and what do you hope to do with this feature? > TIA, > Stefano > > Scrive "James R. Leu" <jl...@mi...>: > > > I spoke too soon, someone is working on it. I will make sure to > > post to the list when I have patches for it. > > > > On Fri, Sep 24, 2004 at 06:16:56PM +0200, Stefano Avallone wrote: > > > Hi all, > > > > > > we are interested in multiple FTN-NHLFE associations. Our goal is to split > > > > > a single flow (identified by means of source and dest address and port, > > > protocol) over different LSPs. According to RFC 3031, the FTN may map a > > > particular label to a set of NHLFEs, but exactly one element of the set > > > must be chosen before the packet is forwarded. Does mpls-linux already > > > contain some code about such functionality? If not, any hint on how to > > > proceed? > > > > > > Thanks a lot in advance, > > > Stefano > > > > > > -- > > > Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/ > > > > > > > > > ------------------------------------------------------- > > > This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 > > > Project Admins to receive an Apple iPod Mini FREE for your judgement on > > > who ports your project to Linux PPC the best. Sponsored by IBM. > > > Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php > > > _______________________________________________ > > > mpls-linux-general mailing list > > > mpl...@li... > > > https://lists.sourceforge.net/lists/listinfo/mpls-linux-general > > > > -- > > James R. Leu > > jl...@mi... > > > > > > > ---------------------------------------------------------------- > This message was sent using IMP, the Internet Messaging Program. > > > ------------------------------------------------------- > This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 > Project Admins to receive an Apple iPod Mini FREE for your judgement on > who ports your project to Linux PPC the best. Sponsored by IBM. > Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php > _______________________________________________ > mpls-linux-general mailing list > mpl...@li... > https://lists.sourceforge.net/lists/listinfo/mpls-linux-general -- James R. Leu jl...@mi... |
|
From: Stefano A. <sta...@un...> - 2004-09-29 16:11:09
|
We hope to be able to split a single flow over multiple LSPs, both for load balancing and security reasons. We are thinking of an iptables rule on the ingress LSR which classifies the packets belonging to the same flow and targets the appropriate entry into the NHLFE table. Such entry can contain an MFWD instruction, which should be of the form: mfwd:mf:se:key1:key2:...:keyN where mf is the multicast flag and se is a sort of "splitting engine". The specification of a splitting engine (via a short name or an id) allows for different splitting approaches. The splitting engine should be a function of the sequence number of the packet and return the id of the selected LSP. The sequence number may be put into the nfmark field by a proper iptables target (I think there does not exist a target that marks the nfmark field with the value of the packet counter) and read by the mpls code. The splitting engine may be as simple as a round robin scheduler or as complex as to account for network resource utilization. In the case of splitting of a whole FEC (instead of a single flow) such approach may be used as well. Any comments? Best regards, Stefano Scrive "James R. Leu" <jl...@mi...>: > On Mon, Sep 27, 2004 at 05:48:28PM +0200, Stefano Avallone wrote: > > Hi James and all, > > > > we agree with your view, we were thinking of an instruction for the NHLFE > table. > > > > May we have more information about the current work? Maybe we can contact > the > > authors to exchange views/contribute to the work,... > > Christophe Fillot <cf...@ut...> is working on the code and has written > some operational code in kind of a 'proof of concept' form. He and I > discussed what the possible next steps are. Maybe your input could help > influence this process. What are you looking for, and what do you hope > to do with this feature? > > > > TIA, > > Stefano > > > > Scrive "James R. Leu" <jl...@mi...>: > > > > > I spoke too soon, someone is working on it. I will make sure to > > > post to the list when I have patches for it. > > > > > > On Fri, Sep 24, 2004 at 06:16:56PM +0200, Stefano Avallone wrote: > > > > Hi all, > > > > > > > > we are interested in multiple FTN-NHLFE associations. Our goal is to > split > > > > > > > a single flow (identified by means of source and dest address and port, > > > > > protocol) over different LSPs. According to RFC 3031, the FTN may map a > > > > > particular label to a set of NHLFEs, but exactly one element of the set > > > > > must be chosen before the packet is forwarded. Does mpls-linux already > > > > > contain some code about such functionality? If not, any hint on how to > > > > > proceed? > > > > > > > > Thanks a lot in advance, > > > > Stefano > > > > > > > > -- > > > > Using M2, Opera's revolutionary e-mail client: > http://www.opera.com/m2/ > > > > > > > > > > > > ------------------------------------------------------- > > > > This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 > > > > Project Admins to receive an Apple iPod Mini FREE for your judgement > on > > > > who ports your project to Linux PPC the best. Sponsored by IBM. > > > > Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php > > > > _______________________________________________ > > > > mpls-linux-general mailing list > > > > mpl...@li... > > > > https://lists.sourceforge.net/lists/listinfo/mpls-linux-general > > > > > > -- > > > James R. Leu > > > jl...@mi... > > > > > > > > > > > > > ---------------------------------------------------------------- > > This message was sent using IMP, the Internet Messaging Program. > > > > > > ------------------------------------------------------- > > This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 > > Project Admins to receive an Apple iPod Mini FREE for your judgement on > > who ports your project to Linux PPC the best. Sponsored by IBM. > > Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php > > _______________________________________________ > > mpls-linux-general mailing list > > mpl...@li... > > https://lists.sourceforge.net/lists/listinfo/mpls-linux-general > > -- > James R. Leu > jl...@mi... > ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. |
Thanks for helping keep SourceForge clean.
X