Re: Does it matter where the binaries are?

my friend said:
Server mount the storage using CFS.
Diffrent servers have diffrent SIDs and same ORACLE_HOME,the spfile / password-file would not have any conflictation.

Our system is implemented with exactly this configuration.(Surely,we consult metalink before impliment)
Smoothly run for 1 year.

On Thu, 10 Mar 2005 23:13:15 +0000, stephen booth <stephenbooth.uk_at_gmail.com> wrote:
> On Thu, 10 Mar 2005 14:27:41 -0800 (PST), Jeremiah Wilton
> <jeremiah_at_ora-600.net> wrote:
> > People die if your system goes down?
> >
> > Please tell me you're not with the NHS :-)
>
> Close, city council.
>
> Here's a scenario. Social worker believes that a child is in danger
> but before they can remove that child they need to get a court order
> signed by Justice of the Peace. Before the JP can/will sign the
> order they need to be shown certain pieces of evidence. The paper
> copies of that evidence might be in 3 or more physical locations which
> may be closed up for the night and may require special permissions to
> access (e.g. documents held by the NHS). If we have copies of those
> documents in a DMS then we can print copies and the social worker can
> get the court order. If the DMS is down then by the time the paper
> copies can be assembled and taken to the JP an abuser could have
> killed the kid or absconded with them. The next day you're reading in
> the newspaper about how the city council didn't do enough to protect
> [insert name here].
>
> That's not a hypothetical, it's happened. That's what I mean by
> people dying if the system is down.
>
> >
> > I think "if a system goes down" is the key phrase here. What if that
> > system is the filer? Or the network between the hosts and the filer?
> > Or the host network adapter?
>
> According to NetApp that would never happen.
>
> Yeah, I know they're being less than 100% accurate. Unfortunately
> project managers in the public sector tend not to be chosen for their
> ability to think.
>
> We've managed to get them to look at a clustered filer so that at
> least if the network interface or one of the filer heads dies we'll
> still have a service. Of course that doesn't protect us from power
> outage, network switch failure or a JCB driver getting over
> enthusiastic and cutting you're nice fat pipe (it sounds impossible
> until it happens to you). My personal preference would be for two
> filers, with their attendant server(s), on separate sites where each
> database stored on one filer was paired with a standby stored on the
> other. I'm hoping that once we get the system bedded in and start
> getting more services online that I'll be able to get something like
> that.
>
> >
> > Will the project manager compensate the victims' families for deaths
> > resulting from poor availability design?
>
> Project managers can be rather like seagulls (appologies to any
> project manager's offended by this, maybe I've become jaded by my
> experiences). The have the ability to fly in, make a lot of noise and
> spread their guano then fly out and onto the next project. The rest
> of us have to live with the results of their decisions.
>
> >
> > I apologize if this sounds too critical. You are right for asking
> > these questions here.
>
> I only ask questions when I want an answer. A lot of this is my
> trying to produce something usable whilst operating within the
> parameters set for me by the project managers.
>
> Stephen
>
> --
> It's better to ask a silly question than to make a silly assumption.
> --
> http://www.freelists.org/webpage/oracle-l
>

-- 
Kamus <kamusis_at_gmail.com>

A Oracle8i & 9i Certified DBA from China
--
http://www.freelists.org/webpage/oracle-l