Re: Bi-Directional replication

From: Vadim Keylis <vkeylis2009_at_gmail.com>
Date: Wed, 28 Sep 2016 11:15:01 -0700
Message-ID: <CAHL4c1MErxdU-kQEvm0WLwU9fMgXvA3wCW2-b0CzR03M=DXzXw_at_mail.gmail.com>



Good morning. Thanks so much for valuable advise that I need to sort through. I am sure I will have more question.

Thanks so much again.

On Wed, Sep 28, 2016 at 8:04 AM, Yong Huang <yong321_at_yahoo.com> wrote:

> > - Don't use Table-based sequences (e.g. for a no-gap sequence): you
> can't have no-gap sequences.
> > - If you are using SEQUENCEs, don't replicate them but ensure they can't
> produce the same value in each DB.
> ...
>
> Since sequences pose a great challenge, maybe GUID as given by the
> sys_guid() function is a good option? Theoretically, the raw values from
> this function will never be repeated by any database. Of course this won't
> solve the problem that the same record (same from the business point of
> view) may be inserted twice.
>
> Instead of bi-directional replication, if budget is not an issue and the
> two data centers are not too far apart, say less than 100 km, extended RAC
> can also be considered.
>
> Yong Huang
>

--
http://www.freelists.org/webpage/oracle-l
Received on Wed Sep 28 2016 - 20:15:01 CEST

Original text of this message