Re: Data Guard (zero data loss)

In article <3e6bb9a2$1_1_at_mk-nntp-1.news.uk.worldonline.com>, "Paul says...
>
>"Timmy Sin" <tswsin_at_yahoo.com> wrote in message
>news:b49nof$dg3$1_at_news.hk.linkage.net...
>> Hi All,
>>
>> Does anyone have experience on the zero data loss feature with Data Guard
>> 9i? It states that modification would be committed only when the redo
>> information has been at least transferred to one of the standby databases.
>> Is this feature practical in real life? How about if the primary and
>standby
>> databases are connected by a T1 WAN link?
>>
>> Thanks,
>> Timmy
>>
>Surely this requires no more than common sense? The technical mechanism is
>irrelevant; if one is going to guarantee no data loss, then there has to be
>a successful save in two places before the transaction is committed.
>
>Depending on the degree of resilience and recoverability required, one
>simply evaluates the alternatives available; and then makes a decision.
>
>Regards,
>Paul
>
>

Absolutely spot on, Paul. Particularly this bit - "Depending on the degree of resilience and recoverability required". There's so many people out there who *think* they need soemthing and therefore do it, rather than building a business justification for it first. I still remember well having to do an OPS install for a site that had a "mission critical" application. No choice, just do it. When I got to the site, their mission critical app had one table, 40,000 rows. THAT WAS IT! THese people had bought a cluster and OPS to support something that could run on a spreadsheet! Oh, and when I questioned them on their downtime capacity, nobody was actually using the app yet. This is what they *thought* the requirement might be. Oi vay!

HTH. Additions and corrections welcome.

Pete

SELECT standard_disclaimer, witty_remark FROM company_requirements; Received on Mon Mar 10 2003 - 12:25:14 CST