DSXchange

I don't think so. The topology of a specific installation (not project) is part of the output of an ISALite general health check. But you need administrator privileges to run that.

priyadarshikunal wrote:Unless entire is used on all links and running on more than 1 partition.

... which is, of course, one of the most elegant ways to ruin the result of almost every DataStage parallel job...

If your primary-key is an incremented surrogate-key from a sequence or a key-file you can also keep record of the highest key used before load and delete all records with a higher value in case of faillure.

This should be quick, because the primary-key is used to identify the rows to be deleted.

Either 1) have a LOAD_ID that is incremented at every load. Delete only the rows for the last LOAD_ID (have an index on the LOAD_ID-Column or - in some cases even better - partition your table by LOAD_ID) Or 2) Save the Timestamp of insertion of a new record in the target-database and delete only re...

Don't know why you name this thread "Inner Join" and ask a question concerning left outer afterwards, but be that as it is. Left outer join outputs all rows from the left (main) stream and combines them with all rows matching on the key-column[s] from all other input-links. Duplicate key-v...

Talking about RCP requires a basic understanding of the way DataStage works under the covers. The Sort-Stage is basically just a more comfortable GUI for the OSH-operator tsort with improved visibility on the canvas provided by its Icon. But the GUI forces you to select a key-column from a list of c...

So, what was the solution to solve this?

Hi Jerome,

Yes, sure it is. I just doubt your performance gains will be all that significant in a job You are just going to use during a certain test-period.

Hi rohit, this depends on the level of security You wish to have. It is certainly possible to use the same dsadm-user for both installations. But is it also advisable? Choosing different users is more secure, because Your admin-user for one installation has got no privileges to make any changes on t...

Hi Jerome, the important question is: what do you want to achieve by comparing CRC32 of two data-rows instead of comparing the column values themselves? You'd still have to read all rows and all columns from both extracts and to generate the Checksum for both. So you do not reduce the amount of data...

Yes, it is possible. You have to be very careful not to use any server-ports for both installations at the same time. IBM advises to contact the concierge-team for this type of installation. They provide valuable information on necessary settings for parallel installation of two versions of DataStag...

Which User do the folders under C:/IBM/InformationServer/ belong to?

This is the user you are looking for. The important thing is - as mentioned - write-access to the file-system.

Delete (by script in before-Sql) then insert does not sound sensible to me, since it involves running delete first and then insert afterwards , while your updates would probably take about the same time as the deletes (they involve the same index-search). Just used "upsert" as a short form...

or Does this mean we need to create a new Windows Domain User (like a new process id) with a non expiring password..add this user in the Information Server and then map its credentials accordingly (one time only)? No. You need either the datastage-adminuser (usually dsadm) or a user belonging to th...

Hi Tony, One problem with Upsert is that You have to set the array-size to 1 on the connector, so it will definitely slow down considerably. I would opt for a left-join-operation against the target-table to identify the rows to be updated and the rows to be inserted. If there is a chance that a lot ...