Comments on: Distributed Pipelines

By: Sayan Malakshinov

Sayan Malakshinov — Wed, 17 Oct 2012 16:04:46 +0000

It is much better to make a “rollback”, and in fact instead of “commit” executed “rollback” on the remote database. I wrote about in the Oracle-L mailing list and in my blog: http://orasql.org/2012/10/13/deceptive-commit-after-select-from-dblink/

By: sky_lt

sky_lt — Tue, 17 Apr 2012 13:58:02 +0000

It is possible to avoid ORA-12840: cannot access a remote table after parallel/insert direct load txn.

But we have to include pragma for autonomous_transaction and to commit inside pipeline function.

By: Partitioned Bitmaps « Oracle Scratchpad

Partitioned Bitmaps « Oracle Scratchpad — Fri, 01 Jul 2011 17:21:28 +0000

[...] it would be pretty easy to write some sort of solution using pl/sql and pipelined functions – perhaps a function that takes a table_name loops through each partition of the table in turn [...]

By: John Seaman

John Seaman — Thu, 14 Oct 2010 03:50:00 +0000

I fully appreciate the role office politics can play! I just wanted to make sure I hadn’t missed something from the technology point of view. I hadn’t thought of a distributed view on the remote database either so I did miss something :-)

By: Jonathan Lewis

Jonathan Lewis — Wed, 13 Oct 2010 07:47:08 +0000

John,

That would work – but sometimes you’re not allowed to push data into “someone else’s” database, even if you all work for the same company.

A similar, but less “invasive” strategy would have been to create a distributed join view at the remote database and insert the contents of the view.

Sometimes the choice isn’t about technology, it’s about perceived risks (or politics, or standards).

By: John Seaman

John Seaman — Wed, 13 Oct 2010 03:37:24 +0000

I’m probably missing something but wouldn’t it have been just as easy to copy the small table to the remote node and run the query there?

By: Jonathan Lewis

Jonathan Lewis — Sat, 09 Oct 2010 08:35:00 +0000

Bernard,

The key expression is: “connection qualifier”.

As a matter of routine I have the parameter “global_names” set to true, which has the description: “enforce that database links have same name as remote database”

But to allow more than one db_link to point to the same database the links can be given a qualifiers – which is the extra “‘@” part.

tnsnames.ora entry:

D11G = 
	(DESCRIPTION = 
		(ADDRESS = 
			(PROTOCOL = TCP)
			(HOST = HP-LaptopV1)
			(PORT = 1521)
		)
		(CONNECT_DATA =
			(SERVER = DEDICATED)
			(SERVICE_NAME = d11g)
		)
	)

Code to define of database link:

create public database link d11g@loopback using 'd11g';

By: Jonathan Lewis

Jonathan Lewis — Sat, 09 Oct 2010 08:26:41 +0000

msosar,

No, not as far as I know.

By: msosar

msosar — Fri, 08 Oct 2010 22:06:41 +0000

if we use direct path insert with append hint, this result in:
ORA-12840: cannot access a remote table after parallel/insert direct load txn.

can we avoid this?

By: bernard polarski

bernard polarski — Fri, 08 Oct 2010 14:01:34 +0000

It is not easy to figure out what is behind this ” t2@d11g@loopback t2″.
can you print the relevant part in the tnsnames.ora for ‘d11g’ and ‘loopback’ ?