DSXchange

And the best answer is "Why?" The entire philosophy in parallel jobs is "one task one stage type". If you want to sort make it obvious to future developers, by using an explicit Sort stage. There are two other ways within the ETL tool, and you can also use external methods. Try a...

ag_ram wrote:Please provide a solution to this problem.

Aren't you paying an official support provider to do just this?

For a start, it's hardware. It's a BladeCenter platform running Red Hat Linux, with Information Server (with DataStage and QualityStage and Tivoli Load Leveler preconfigured) with three blades. As you need more capacity you can slot in up to 11 more blades. A search on IBM website for "Informat...

Attention to detail is one of the hallmarks of a professional ETL developer; one can not afford mistakes, particularly preventable ones. This includes laziness in documentation, both at work and here on DSXchange.

Posting in the correct forum (Site/Forum) would be a good start.

This forum (FAQ Discussion) is to discuss topics that would make a good FAQ topic in the FAQ forum.

Your code will only work once, because you overwrite the variable called file with the record you read from the hashed file. This is a Bad Thing. Use a different variable name as the target of the read.

... short of killing the Oracle session.

You're right. I really liked the blade server solution that I saw at IOD. I've been waiting for an "ETL Appliance" ever since the "Data Warehouse Appliance" became popular. Unfortunately, most large organizations still haven't caught on... they still think they need to spend mil...

You can not read a record from a hashed file without the primary key. The primary key is used by the hashing algorithm to determine the physical storage location of the record. The Read statement reads the entire record. You need to know, perhaps from the metadata (the file dictionary), which field ...

No.

And the answer is, and remains, not in server jobs.

Parallel jobs give more flexibility, through the use of schema files, but things are always metadata driven.

Just don't expect any "T" in your "ETL".

You can't get that information from that "report". You need to either farm the logs or query the appropriate links directly (for example using DSGetLinkInfo() function or dsjob -linkinfo command) for their counts.

You can decompose these VarChar columns into day, month and year columns then re-organize them into YYYY-MM-DD format strings, which may be compared directly as strings. Don't forget to add a leading zero to each day and month number that is a single digit.

Job name is the key column in DS_JOBS, so you need it (possibly in a variable) before you can read that row anyway. You are on the right track with the Open statement, but then you need to use Read statements to read the records from the hashed file. Both of these can be found in the DataStage BASIC...