Parallel option in Teradata Mload stage

Post questions here relative to DataStage Enterprise/PX Edition for such areas as Parallel job design, Parallel datasets, BuildOps, Wrappers, etc.

Moderators: chulett, rschirm, roy

Post Reply
sateesh
Premium Member
Premium Member
Posts: 6
Joined: Fri Oct 19, 2007 10:34 am

Parallel option in Teradata Mload stage

Post by sateesh »

When I try to change in advance tab from sequential to parallel in TD Mload stage the job is failing.
Error: Check MultiLoad report for details.
When i look at report it has stopped at Multiload Initial Phase.
Is there anyway we can make it work?

Thanks in Advance.
rcanaran
Premium Member
Premium Member
Posts: 64
Joined: Wed Jun 14, 2006 3:51 pm
Location: CANADA

Post by rcanaran »

Craig, Arnd and Ray, correct me if I'm wrong, but,

MLOAD locks the table for UPDATE. When the MLOAD stage from DS tries to load multiple parallel streams, it tries to run one MLOAD per (node?). 1st one gets successful lock, all subsequent nodes/data streams fail.

Last I worked on a Teradata target from DataStage, the only choices were to change the MLOAD to sequential mode OR limit the number of nodes to ONE.

As far as I remember, TERADATA wants a single datastream for MLOAD and IT will parallelize the datastream across the AMPs based on the Primary Index. Accepting parallel datastreams from another product may interfere with Teradata's ability to optimize the load.

You could try using a TPUMP instead if you really want to run parallel, but I suspect that this will actually be SLOWER and you still run the risk of a HASH/AMP level lock (similar to row-level, but not quite).
keshav0307
Premium Member
Premium Member
Posts: 783
Joined: Mon Jan 16, 2006 10:17 pm
Location: Sydney, Australia

Post by keshav0307 »

because you can not run more then one MLOAD on a table at a time..
making it parallel means more then one MLOAD will trigger on the target table.
sateesh
Premium Member
Premium Member
Posts: 6
Joined: Fri Oct 19, 2007 10:34 am

Post by sateesh »

keshav0307 wrote:because you can not run more then one MLOAD on a table at a time..
making it parallel means more then one MLOAD will trigger on the target table.
The answer is not convincing.
If that is the case why we have even option out there?
I check the Teradata Enterprise stage it has the option too but it is working fine when i configure to parallel.
I am just trying to understand not to urgue.
Thank you.
keshav0307
Premium Member
Premium Member
Posts: 783
Joined: Mon Jan 16, 2006 10:17 pm
Location: Sydney, Australia

Post by keshav0307 »

If that is the case why we have even option out there?
check with IBM.

the same stage(MLOAD) is used for TPUMP also which can run in parallel.

Teradata Enterprise stage is meant for running in parallel, it uses Fastload/fastexport
Post Reply