Re: Goldengate and DW

GG and Dataguard are completely different products for different purposes. So I'm not going to comment on that.

But in case of using GG for data warehouse (in order to extract subset of data and distribute them to Kafka or other databases) I have a few comments. All assuming "data warehouse" is a database that has long running transactions, modifying huge number of rows in the same transaction. Potentially also developers connecting directly to the database and modifying data there and... forgetting to close their transactions for days.

• Memory usage for extract can get pretty huge, 50GB+ for our case
• Long running transactions are the enemy of extract, if you need add a new extract, it must go back to the beginning of oldest currently running transaction and start processing from that point. If you stop extract normally and start again, it needs to go back to the bounded recovery checkpoint and start processing from that. If database has huge amount of changes, all that takes a lot of time (hours in our case).
• If replicat encounters a huge transaction (millions of rows), applying it will take a long time, while all other transactions just wait

On Mon, Jan 13, 2020 at 8:18 PM Ram Raman <veeeraman_at_gmail.com> wrote:

> Listers,
>
> Can we consider GG (goldengate) for replication of our multi TB data
> warehouse. Anything nuances, surprises there? Between GG and Dataguard
> which one can be a better product.
> Licensing cost is NOT an issue.
>
> Thanks
> Ram
> --
>
>
>

-- 
Ilmar Kerm

--
http://www.freelists.org/webpage/oracle-l