Date: Thu, 13 Dec 2007 12:07:14 -0800 From: Julian Elischer <julian@elischer.org> To: Nikos Vassiliadis <nvass@teledomenet.gr> Cc: freebsd-net@freebsd.org, Max Laier <max@love2party.net> Subject: Re: bikeshed for all! Message-ID: <476190F2.2030105@elischer.org> In-Reply-To: <200712131549.21669.nvass@teledomenet.gr> References: <476061FD.8050500@elischer.org> <200712130021.56473.max@love2party.net> <476072DB.3090600@elischer.org> <200712131549.21669.nvass@teledomenet.gr>
next in thread | previous in thread | raw e-mail | index | archive | help
Nikos Vassiliadis wrote:
> On Thursday 13 December 2007 01:46:35 Julian Elischer wrote:
>>> pf has ifdef'ed out code to deal with the OpenBSD version of routing
>>> tables. What it does is adding an mbuf_tag which carries the tableid
>>> and ip_{{in,out}put,forward} take action accordingly.
>> EXACTLY what I plan on doing.
>
> It would be nice to be compatible with the OpenBSD pf.
>
> Or perhaps even the OpenBSD multiple forwarding table infra-
> structure. If I recall correctly other programs in their
> base system - such as OpenBGPD - can interact with it.
In functionality, I have now has at least a passing similarity to what
OpenBSD are doing.. in this version at least, which is what I will
commit first as it is simple enough to back port to 6.x/7.x.
The OpenBSD scheme is extensible by which I mean the number of
tables can be grown dynamically, but in the version I will commit
you must pre-define the number at boot time (with a default set at
compile time)
Extensibilty will break some ABIs and as such will have to be in 8.x only.
I plan on doing several of the internal mechanics different from openBSD
so I do expect a small divergence, but hopefully not enough to make
a large difference to the user. I will also add a -T option to tools such
as netstat and route, so you don't have to use the separate tool with them,
just like in OpenBSD.
Now on to the topic at hand, shed colour...
suggestions have been:
Base short version utility name
==================================================================
instance (ala Juniper) inst? rtinst rtinst
view, netview routeview, rtview setview
zones rtzone rtzone,setzone
vrf (ala cisco) setvrf
vhost vhost
vnet vnet
domain dom rtdom, netdom
plane rtplane, setplane
cloud hmm no, maybe not..
vrf is a good one but I think it may map closer to what vimage is..
you can map interfaces into vrfs and vimages, where multiple routing tables
doesn't map interfaces at all.
I like instances too.. it has a good feel to int.. but shortens to inst which
makes netinst look like netinstall.. rtinst? maybe.
zones sounds like solaris zones which is basically vimage.
plane makes me think of the data plane vs control plane in routers.
possibly one could think of having multiple data planes..
>
> Nikos
> _______________________________________________
> freebsd-net@freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-net
> To unsubscribe, send any mail to "freebsd-net-unsubscribe@freebsd.org"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?476190F2.2030105>