Re: [Bier] BIER in IPv6 --- draft-zhang-bier-bierin6-04

["Xiejingrong (Jingrong)" <[e-mail]>]

Hi Tony,
Thank you for your clarification.
I think I got your points here regarding BIERin6 and why you think there is no need for a new L3-BIER.
Please correct me if I understand anything wrong.

We are _not_ "tunneling" in case of LL to LL BIERin6 in fact for all practical purposes (tunnel assumes fragmentation/security etc which BIER architecture does NOT presume from L2 header, we do not expect Ethernet to fragment or encrypt BIER).
[XJR]
Once IPv6 tunnel is used, then fragmentation/security are big concerns.
BIER architecture was designed for L2.5 and does not presume L2 link for fragmentation/security always.

IPv6 is basically used the same way we use ethernet, it's poor man's "L2 header" that allows standard, non-extended silicon to throw stuff to slow path to process if silicon doesn't support BIER (will be reality for e.g. cheap WiFi routers for ages to come I think).
[XJR]
In BIERin6, the IPv6 is used just like “Ethernet” link, but better for Poor man ---- who want to implement BIER in software.

If silicon does support BIER fast path can be taken as well with bierin6 but if silicon supports BIRE we have L2 encaps which arguably should be used rather than twisting ipv6.
[XJR]
If silicon does support BIER fast path, BIERin6 can be easily implemented in fast path too.
But given that there is already L2.5 BIER encapsulation, argument is that L2.5 may be preferred rather than using BIERin6.

BIER architecture uses unicast tunnels to get frames across parts of network without BIER support as well.
[XJR]
BIER has its architecture to use unicast tunnels to get frames across parts of network without BIER support.

Now, here the bierin6 is irresistibly convienent to just stick a BFR prefix as destination and hop over routers without even signalling a tunnel and then it is a unicast tunnel from BIER perspective.
[XJR]
BIERin6 only need to use IPv6 unicast tunnels with BFR prefix as tunnel endpoint to “get frames across parts of network without BIER support”----without signaling a tunnel and without manually configuring the tunnel-end IPv6 address.

Kind of like SR or SRv6 or MPLS tunnel (which need to be differntatied from BIER running natively over MPLS labels) but for v6 no special HW support is needed so that's neat ;-)
[XJR]
There is other kind of IPv6 unicast tunnel like SR/SRv6/MPLS-P2P-LSP, but BIERin6 is even better(neat) than them.

SR/SRv6/IP forwarding/GRE are just unicast tunnels to BIER.
[XJR]
SR/SRv6/IP-tunnel/GRE-tunnel are just unicast tunnels to BIER.

[XJR] So comes to the point why you may think there is no need for a new L3-BIER:
BIER was a L2.5 design from its beginning (back to 6 years before) and should not be extended for L3.
BIER has a perfect architecture based on this L2.5 design, and has a perfect architecture already to support “getting frames across parts of network without BIER support” by just using a unicast tunnel of any type.

So in your opinion there is no need for a new L3-BIER. Am I understanding right ?

If so, I can focus on this problem and give my points why a L3-BIER is needed (and of course why the L2.5-BIER can’t work well) for our use cases.

Thanks and best regards,
Jingrong


From: ipv6 [mailto:[e-mail]] On Behalf Of Tony Przygienda
Sent: Friday, March 27, 2020 12:11 AM
To: zhang.zheng <[e-mail]>
Cc: BIER WG <[e-mail]>; 6MAN <[e-mail]>
Subject: Re: BIER in IPv6 --- draft-zhang-bier-bierin6-04


Yes, critical point. We are _not_ "tunneling" in case of LL to LL BIERin6 in fact for all practical purposes (tunnel assumes fragmentation/security etc which BIER architecture does NOT presume from L2 header, we do not expect Ethernet to fragment or encrypt BIER). IPv6 is basically used the same way we use ethernet, it's poor man's "L2 header" that allows standard, non-extended silicon to throw stuff to slow path to process if silicon doesn't support BIER (will be reality for e.g. cheap WiFi routers for ages to come I think). If silicon does support BIER fast path can be taken as well with bierin6 but if silicon supports BIRE we have L2 encaps which arguably should be used rather than twisting ipv6.

BIER architecture uses unicast tunnels to get frames across parts of network without BIER support as well. Now, here the bierin6 is irresistibly convienent to just stick a BFR prefix as destination and hop over routers without even signalling a tunnel and then it is a unicast tunnel from BIER perspective. Kind of like SR or SRv6 or MPLS tunnel (which need to be differntatied from BIER running natively over MPLS labels) but for v6 no special HW support is needed so that's neat ;-) SR/SRv6/IP forwarding/GRE are just unicast tunnels to BIER.

--- tony