[Ubuntu-PH] Ubuntu Expertise needed

[Justin Jereza]

On Tue, Mar 8, 2011 at 2:45 AM, andrew <andrewuy888 at gmail.com> wrote:
> 1. primary domain controller running openldap, managed by webgui

You can have the Linux boxes authenticate with LDAP directly or with
Kerberos if you're really paranoid.

> 2. nfs and samba server (for the few windows boxes, around 5). home folder
> mounted as a separate partition (need to enforce quotas)
>
> 3. around 25 desktops (home directory pathed to nfs/samba server when they
> log in)
>
> 4. backup server for the pdc and file server and sql server (db is around
> 30gb compressed), must dump to a nas and a tape (offsite backup, manual
> intervention of changing tapes at 8am daily).  backup and restore should be
> easy.

I like Bacula. Difficult to configure but everything is automated once
you're done. All you need to do is swap tapes when you get an e-mail
telling you to do so. Restores are pretty straightforward but since
some functions require issuing commands to the console, ease of use is
relative.

> 5. backup of pdc, files server and backup server (virtualize?) (auto
> failover if available)

You can use DRDB to keep the file server synced. Only issue I can see
are stale file locks when switching to backup. If you have cash to
spend, you can use SANs instead. Two SANs and two VM hosts should be
enough. It will really depend on whether those costs are justifiable.

> 6. users can print to a central fax server (with two usb modems?).  incoming
> faxes are centralized, reviewed and dispatched to users. (auditing for what
> users printed, how many pages, what doc if available, softcopy of all
> records)
>
> 7. one update server for the client desktops (to save bandwith)

debmirror works nicely for this so that you don't have to mirror the
entire repo. As an added bonus, you can do PXE installs of the
workstations as well.

> 8. Print audit, who print, what, pages, with report summary. softcopy of all
> print requests if available

CUPS keeps a log of all prints so you might not even have to do
anything besides process logfiles.

> 9. hardware design? dell r710?
>
> 10. manage, alerts (email when errors occur like system stopped
> unexpectedly...)

I like openNMS better than Nagios. openNMS just polls SNMP data. I
even have it configured to send me an SMS if something goes wrong.

> 11. all processes are automatic
>
>
> security concerns are important.

Just remember that increasing security, reliability, availability, and
automation will increase costs exponentially.

> please feel free to recommend, we're looking for a best practice setup and
> some additional features.  if there are other, then please do recommend.
>
> can you please point me in the right direction for a person that is
> qualified to do this systems infrastructure?
>
>
> thanks!
>
> andrew
>
> --
> ubuntu-ph mailing list
> ubuntu-ph at lists.ubuntu.com
> https://lists.ubuntu.com/mailman/listinfo/ubuntu-ph
>
>



-- 
Justin Jereza
LPIC-1