Where to for Samba4? - Native IPv6

[davidholder at erion.co.uk]

Andrew,

I think it is important that Samba4 should be native IPv6. This is going
to be necessary as the world moves to IPv6. Particularly next year, with
the release of Windows Server 2008 and with the requirement for IPv6
support in Federal agencies in the US.

IMHO it is worth considering dropping support for the historic IPv4 socket
API and only supporting the IPv6 socket API. Use of IPv6 within the Samba4
internals would allow you to use one set of function calls for IPv4 and
IPv6. Essetially, if you code for IPv6 then you get IPv4 for free (this is
little bit of a simplification).

This would bring you into line with Vista and Windows Server 2008 and with
many other platforms and products that are now IPv6 native.

Whether you take this approach or not, IPv6 support should be high on the
list. There are many organisations that have been holding off implementing
IPv6 due to lack of full support in AD, with the release of Windows Server
2008 this will change. Some may drop Samba as a result.

I'm on holiday at present, so I may be slow to reply to any comments! I'll
do my best!

Regards,
David



> Now that we have an alpha release, I thought I would try and start a
> discussion about where Samba4 should go.
>
> Clear targets we already have on the Wiki are:
>  - Clustering support.  This code started in Samba4, and I hope that in
> time, Samba4 can use these features.
>  - Schema enforcement.  One, not unreasonable, suggestion has been to
> make this become the responsibility of an OpenLDAP (or Fedora DS)
> backend.
>  - Access control.  We now store NT ACLs in LDAP, and honouring them
> (rather than kludge_acls) should not be that hard.  Likewise on the
> registry (should have been done for the alpha...)
>  - One-off active Directory migration.
>
> Some other tasks I see for the near future:
>  - Kill the LDB browser and move back to SWAT1 in Samba4, with
> phpldapadmin either run inside SWAT, or with apache.
>
> A list of targets is useless, if they bear no resemblance to what is
> likely to be achieved.  Similarly, most of the work done on Samba4 in
> the past few months has not been on the roadmap, but just the response
> to well-filed bugs in bugzilla.
>
> One particular thought I've had suggested is to try using OpenLDAP to
> handle schema validation and subtree renames (by always using an
> OpenLDAP hdb backend).  Likewise, should we trim unlikely features such
> as NetBIOS browsing from the list of features folks may expect from
> Samba4?
>
> Finally, I've heard all manner of different people give views in the
> press about where Samba4 is heading - a separate release, a DC only,
> just another head of a Samba3 borg?
>
> From my point of view, I'm expecting to make a release, with good DC,
> and hopefully good file-server capabilities.  But I would like to
> discuss what (if anything) Samba4 means to more than just myself.
>
> Thoughts?
>
> Andrew Bartlett
> --
> Andrew Bartlett
> http://samba.org/~abartlet/
> Authentication Developer, Samba Team           http://samba.org
> Samba Developer, Red Hat Inc.
>