RARE user and assistance email list

[mc36 <>]

yess, the decapsulation, routing/switching decirions and encapsulation phases 
are fully independent in all the dataplanes...

On 5/18/22 18:12, Marcos Felipe Schwarz wrote:
> Thanks for the answer,
>
> Just to confirm, these tunnel options you listed can be implemented in-line 
> by RARE FreeRtr, so a PolKA Core FreeRtr P4 device can at the same time, 
> receive a encapsulated PolKA packet (over vlan, vxlan, nvgre, l2tp), 
> decapsulate the tunnel, make routing decisions on the PolKA header, and 
> re-encapsulate before transmiting to the next hop?
>
> A similar approch that is used to transport SR-MPLS packets over legacy 
> networks is encapsulating MPLS traffic is over UDP (RFC 7510).
>
> Regards,
>
> Marcos Schwarz
>
> -----Mensagem original-----
> De: mc36 <>
> Enviada em: quinta-feira, 12 de maio de 2022 16:54
> Para: Marcos Felipe Schwarz <>; ; Cristina 
> Klippel Dominicini <>; Fr  d  ric LOUI 
> <>; Simon Leinen <>
> Cc: Everson Scherrer Borges <>; Magnos Martinello 
> <>; Rafael Silva Guimar  es <>
> Assunto: Re: RES: [RARE-users] Experiments PolKA vs. Segment Routing
>
> hi,
> so it highly depends on the underlay chosen... those that propagate 
> explicitly the upper layer's ethertype, will work..
> examples include vlan, gre and l2tp, all with or without macsec... those that 
> dont, like ipip and wireguard, can only carry
> ipv4 or ipv6 because on the receiving side, one can only have the first 4 
> bits to _guess_ the upper layer protocol...
> br,
> cs
>
>
> On 5/12/22 19:25, Marcos Felipe Schwarz wrote:
> > One question I have.
> > Considering current PolKA Implementation on freeRtr, it can only be run over 
> > a L2 circuit (i.e. a VLAN or direct connection). Or can we run a PolKA tunnel 
> > over a L3 network, similar to VXLAN?
> > I'm asking this, because there are some sites that are able to reach only 10G 
> > over a L2 circuit but have 100G on a shared internet Link. So if freeRtr is 
> > more flexible we can consume more bandwidth for the experiments.
> >
> > Regards,
> >
> > Marcos Schwarz
> >
> > -----Mensagem original-----
> > De: mc36 <>
> > Enviada em: quarta-feira, 11 de maio de 2022 02:13
> > Para: ; Cristina Klippel Dominicini
> > <>; Fr  d  ric LOUI
> > <>; Simon Leinen <>
> > Cc: Marcos Felipe Schwarz <>; Everson Scherrer
> > Borges <>; Magnos Martinello <>;
> > Rafael Silva Guimar  es <>
> > Assunto: Re: [RARE-users] Experiments PolKA vs. Segment Routing
> >
> > hi,
> >
> > On 5/11/22 01:05, Cristina Klippel Dominicini wrote:
> > > Regarding the WCMP feature, if possible, it would be really good for the data 
> > > intensive science use case. In this way, we could load balance between a set 
> > > of tunnels.
> >
> > wcmp was added to the todo.txt and after the the test runs, it's in the todo 
> > too:
> > https://github.com/rare-freertr/freeRtr/commit/dfcaeb0e5cb085434fa35fc
> > 84aa2540b2197f397 until that, you can use multiple tunnels and ecmp...
> > so if you want 1/3-2/3 over 2 paths, then you'll set up 3 tunnels, 1 over 1/3 
> > path, and 2 over 2/3 path...
> >
> > br,
> > cs