Re: Oracle2PostgreSQL Migration with PL/pgSQL

Guido Stepken <stepken_at_little-idiot.de> wrote in news:b4j633$op$05$1_at_news.t-online.com:

> Describing a certain bug always requires mentioning the exact version
> of
> software and code.
> I have heard of mysql 3.22+ crashing, seen oracle 8i and 9i crashing
> on ibm aix with raid hardware (ibms fault). There is always a way to
> crash any database, IMHO.
> Nevertheless there are installations with > 1-2 Terabyte out there,
> based on oracle as well as on postgreSQL 7.1+, sapdb, firebird (former
> interbase...). Postgresql IMHO is the only free database, which is
> suited to replace oracle, due to its pl/SQL compatibility.
> Being delivered with source makes it much easier to debug and correct
> bugs in case of a crash.
>
> tnx to all, who have sent code, hints, experiences to my adress at
> stepken @ little-idiot.de ....
>
> i am quite optimistic, that the migration handbook will soon do very
> serious harm to oracle's sales volumes......my mysql handbook (sorry,
> only in german, still, published in 1999) did it's purpose very
> well.....mysql is #1 in germany now for small databases.....
>
> Oracle changed its sales policy.....resellers can offer oracle for
> about 1.600-1.800 Euros/US$ each......what a price reduction, compared
> to cpu licenses.....the only condition is, that oracle must be sold
> together with a application....
>
> regards, Guido Stepken
>
> Volker Hetzer wrote:
>

>>>PostgreSQL IMHO more and more becomes a Oracle killer. Worth, writing
>>>a small book about it ;-)
>> 
>> Last time I tried I crashed it twice in one day. Seems, postgres'
>> response to faulty sql is not an error message but a crash with data
>> corruption. 

Has anyone seen the article about danske bank losing a week of trading when IBM were upgrading thier DB2 systems. No crash - just grand corruption of the database when batch processes started...

See http://www.theregister.co.uk/content/53/30095.html for more info.

Basically when you scale up even with the best commercial RDBMS you hit problems - SS only support one (maybe two makes of raid array) other raid and NTFS controllers cause SS to chars or corrupt data.

Oracle had major problems with early Sun RAID arrays some years back but this was found and fixed before the relevant project went live.

What happended to IBM/DB2 is not a one off - it happens to all vendors.

What I find scary if that the bank allowed a vendor to upgrade thier **live** systems with a hardware upgrade. When we ship updates to customers we have a (identical as possible) dev system (often the customers - sometimes our own kit) to run the updated system on before even touching the live systems.

I would have expected a Bank to have required new hardware to undergo a number of tests on thier own dev/backup systems - they must have backup failover hardware/systems? - before commiting to upgrades of thier live systems.

Jacqui Received on Fri Apr 04 2003 - 08:27:36 CST