Re: "All triggers are evil",..., really?

  • From: "Niall Litchfield" <niall.litchfield@xxxxxxxxx>
  • To: dbvision@xxxxxxxxxxxx
  • Date: Thu, 4 Sep 2008 12:50:41 +0100

I'm fairly sure that I'd react in much the same way, but actually I
don't think it unreasonable to define availability in terms of
availability within specific windows, esepcially when talking about an
SLA. When I go to Tescos (grocery store), I expect it to be 100%
available within it's advertised hours. I don't expect it to be open
when it says it will be shut. An IT service it seems to me could learn
from this. The kicker of course is when the business decides that the
advent of the web means that you are open all day every day, as you
say pretty damn expensive, probably impossible, and likely not worth
it. But the principle of saying "We have 5 9's availability 8-8 every
day" doesn't seem to me to be a bad thing - Gartner will never go for
it obviously.

Niall

On Thu, Sep 4, 2008 at 11:24 AM, Nuno Souto <dbvision@xxxxxxxxxxxx> wrote:
> Connor McDonald wrote,on my timestamp of 3/09/2008 11:48 PM:
>
>
>> I always find it hilarious when you have to wear the day-to-day hit of
>> capturing changes to every row on every row due to some "cast hard in
>> concrete business rule of - thou shalt audit everything" ....but then take a
>> system outage because a large scale data change has to occur, and the same
>> people then insert .../"but we don't want to audit that/" comes along the
>> line...
>
>
>
> Heh!  This one reminded me of a meeting late last year to
> discuss SLA's for high availability.
>
> me: You want what?
>
> HA"expert": 4 9s availability!
>
> me: As in 99.99%?
>
> HA"expert": No, as in 99.9999%!
>
> me: Are you aware how long that is? In the order of seconds outage per year!
> Veeeeery expensive to achieve!
>
> HA"expert":Yes, but senior management has a report where Gartner says
> companies
> of the size of yours MUST have that level of HA.
>
> me(with instant rash, after "Gartner says"): But how do we maintain and
> patch up our systems without implementing everything in active-passive
> clustering?
>
> HA"expert":: Business has agreed with management there will be a one day
> monthly outage for maintenance!
>
> me: Well, if there is a one day monthly outage then we can't possibly be
> doing 4 9s HA according to your own definition! That's not even 99.99% HA!
>
> HA"expert": Of course we are!  On the rest of the month, we have 4 9s HA!
>
> me: Listen! 4 9s HA is actually, by ANY definition of the IT industry
> *including* Gartner's and the IEEE, 99.99%.  Which already means LESS THAN 1
> HOUR OUTAGE in the WHOLE year! You simply cannot claim 12 days of outage per
> year is 99.9999% HA, that is a FALSE statement! Furthermore: do you realize
> that in our CURRENT configuration we actually achieved 100% (no failures
> whatsoever) HA in this last year? Why do you want to tamper with that?
>
> HA"expert": You are just being intentionally difficult and trying to stop
> this company evolving to this millenium! I am going to report you to senior
> management!
>
>
> Elbonia is real, folks...
>
>
>
>
> PS: the 99.9999% "HA" is dead in the water after they looked at the bottom
> line needed to achieve it!  Sanity prevailed, thank Codd!
>
> --
> Cheers
> Nuno Souto
> in sunny Sydney, Australia
> dbvision@xxxxxxxxxxxx
> --
> //www.freelists.org/webpage/oracle-l
>
>
>



-- 
Niall Litchfield
Oracle DBA
http://www.orawin.info
--
//www.freelists.org/webpage/oracle-l


Other related posts: