Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Multimaster replication confusion!

Re: Multimaster replication confusion!

From: DA Morgan <damorgan_at_psoug.org>
Date: Fri, 03 Jun 2005 12:57:46 -0700
Message-ID: <1117828543.397180@yasure>


julius.cooper_at_gmail.com wrote:
> Daniel,
>
> Thx for the reply.
>
> As for the version: it's Oracle9i (sorry).
>
> Now, as to why we'd want to do it this way? The short answer is that
> we need high availability (99.9999999+), and they want to have local
> store copies of the data at each workstation to ensure that.
> Basically, the situation we're in is this: if a particular transaction
> cannot or does not complete, the entire batch will fail--and we'll get
> hit with huge fines.
>
> Hope that helps.
>
>
> Regards,
> Julius

I've got some very bad news for you. What you are doing will not get you anwhere near that level of availability. So first lets kill the myth.

1 day = 86,400 seconds
1 year = 31,556,736 seconds
0.001% loss of service (99.999) = 316 seconds 0.00001% loss of service (99.99999) = 3.16 seconds and you think you are going to run with 0.03 seconds of down time with replication to multiple workstations?

Two achieve 99.999% requires two diesel generators. Two doors to the data center.
Two independent power conditioning units. Two independent power UPS's.
Two completely independent and redundant networks. Shall I go on?

So you don't have what you think you have and you are totally at risk of even the simplest of problems. If you are seriously in need, for regulatory reasons, of high availability contact me off-line and I will put you in touch with someone that knows how to implement it.

-- 
Daniel A. Morgan
http://www.psoug.org
damorgan_at_x.washington.edu
(replace x with u to respond)
Received on Fri Jun 03 2005 - 14:57:46 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US