Re: Postgresql replication

From: Chris Browne <cbbrowne(at)acm(dot)org>
To: pgsql-general(at)postgresql(dot)org
Subject: Re: Postgresql replication
Date: 2005-08-28 03:35:52
Message-ID: 60oe7iwx5z.fsf@dba2.int.libertyrms.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

William Yu <wyu(at)talisys(dot)com> writes:
> Chris Browne wrote:
>> I'm pretty sure that they _don't_ track balance updates for each
>> transaction that applies to a customer's account. You could, via one
>> form of trickery or another, "overdraw" your account by a fairly hefty
>> amount, and they probably won't notice for a day or even three. But
>> once they notice/estimate that the Gentle Caller has built up some
>> dangerously high balance, they'll warn of impending discontinuation of
>> service if some sum isn't paid by some date.
>
> This works for companies that have some degree of power over their
> customers. E.g. pay up or we disconnect your service. Return your
> overdrafts/pay your fees or we mess up your credit.
>
> This doesn't work if it's a small company who's account has been
> emptied. Sure the bank will refuse to honor the check but then that
> company will be hit with overdraw penalties and possible legal
> penalties to the payee for the bounced check.
>
> The warning threshold system is easy to implement but there will
> always be corner cases where the warning is not soon enough or a
> single payment wipes out ~ 100% of the account. Warn too often and
> it'll be ignored by people as a "boy crying wolf" alarm.

In a context where there is no "degree of power over their customers,"
I would hardly think that the presence/lack of automated controls or
presence/lack of balance synchronization is particularly material.

In other words, if trying to apply policies is going to forcibly
break, then building the data system may have been a futile exercise
in the first place. And trying to distribute the system may again be
a *business* mistake that admits no technical solution.

If that's the case, then we can conclude that replication is no
solution, and that the organization should improve connectivity for
their single centralized system.

That may be the case, but is the uninteresting case, as it amounts to
throwing our hands up, saying "no answers; you can't have a
distributed system," and giving up on trying anything further. No
point to further discussion...
--
let name="cbbrowne" and tld="acm.org" in String.concat "@" [name;tld];;
http://cbbrowne.com/info/linux.html
"There is no reason anyone would want a computer in their home".
-- Ken Olson, Pres. and founder of Digital Equipment Corp. 1977

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Ian Harding 2005-08-28 05:12:51 Re: POSS. FEATURE REQ: "Dynamic" Views
Previous Message Bruno Wolff III 2005-08-28 02:49:29 Re: ?^???G Re: A strange problem