Network configuration future
dcbw at redhat.com
Fri Sep 7 18:17:40 UTC 2012
On Mon, 2012-09-03 at 09:20 +0200, Olaf Kirch wrote:
> Hi Stephen,
> On Wednesday 29 August 2012 19:48:29 Stephen Gallagher wrote:
> > Olaf, I'm very interested to learn more about wicked. Can you perhaps
> > itemize the set of features available to wicked (current as well as
> > in-development) that aren't available (or maybe "accessible) in Network
> > Manager?
> Hm, let me see - this is going to be a bit tough :)
> Currently, wicked supports wired Ethernet, VLANs, bridges, bonds, wireless
> (no EAP mode yet, though), openvpn, and to some degree GSM modems. For
> most of these, the set of properties accessible exceeds those of
> NetworkManager by far.
Can you describe in what way it exceeds?
> In terms of address configuration, it supports dhcp4 and dhcp6, as well as
> IPv4 zeroconf (in addition to static and ipv6 autoconf). You can mix addrconf
> modes (eg configure a NIC with DHCPv4 plus two static routes).
Seems comparable to NM here.
> The layered approach tries to decouple the settings of each layer. For
> instance, you can take down the DHCPv6 agent and bring up the DHCPv4
> agent on an interface without having to cycle it completely. Same for
> adding a bridge port, or removing one NIC from a bond, or for changing
> the offload parameters on an Eth device.
Interesting; something we're also looking at (at least at lower levels)
for NetworkManager, but it also makes it hard to implement a
connection-oriented architecture. NM has had decoupled IPv6 and IPv4
for quite a while now too, actually.
> > I'm involved in a project that's aiming to improve central management of
> > networking and having a comprehensive network stack wrapped in a D-BUS
> > API sounds almost too good to be true from that perspective. A CIM
> > wrapper around such an API could go a long way in that direction.
> Yes, I had been thinking of something like this as well.
> > Do you have a D-BUS API specification available somewhere in a readable
> > form? That would be very helpful.
> Hm, not "readable" in the sense of nice HTML. There's a schema definition
> in the source code which hopefully goes a long way to give you an idea of
> how it works.
> Neo didn't bring down the Matrix. SOA did. (soafacts.com)
> Olaf Kirch - Director SUSE Linux Enterprise; R&D (okir at suse.com)
> SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
> GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg)
More information about the devel