Improve performance

chulett · Post by **chulett** » Fri Jul 31, 2009 5:46 am

OK... I'm going to ask again. Is there anyone there you are working with onsite that knows the Server product and whom can help you with this? Sorry, but you are asking very basic questions on very basic issues and people here are really not meant to be a substitute for proper training. Not to mention that these kind of lessons are best delivered in person, hands on, in an interactive manner.

And I for one don't have time to type up a whole "Hashed Files 101" lesson right at the moment... or any other time, really.

Vishal1982 · Post by **Vishal1982** » Fri Jul 31, 2009 6:41 am

Okay,just let me know how to pass multiple rows to a reference link if my reference input is hashed file.

Sainath.Srinivasan · Post by **Sainath.Srinivasan** » Fri Jul 31, 2009 6:44 am

Hashed files are keyed by id and cannot return multiple rows.

All rows from your input will be passed to your reference unless you mention any constraint on the flow.

chulett · Post by **chulett** » Fri Jul 31, 2009 7:39 am

You don't "pass multiple rows" to a reference link and a hashed file stage cannot return more than one row per lookup - both are singleton operations mandated by the "destructive overwrite" nature of a hashed file which does not allow duplicates per key. If you really need to return multiple rows then you must use either an ODBC stage or a UV stage (which can sit "on top of" a hashed file) and is why I asked the question earlier on.

SHARAD123 · Post by **SHARAD123** » Sun Aug 02, 2009 9:21 pm

Hi Vishal,

To make multiple rows to the hashed file use composite key. But if the requirement is to have duplicate data in the hash file (which I presume will not be the case) then hash file cannot be of use.

Also in the previous posts, there is a mention of not using keys in the lookups. Wondering how can lookup be done without using key columns

. Even if there is an use of Stage variable, only the first row is going to be compared always

chulett · Post by **chulett** » Sun Aug 02, 2009 9:38 pm

There have been times when I've built a reference lookup without keys, but it is rare and it can't be done against a hashed file. And it only really makes sense when you either need to return a static value (and then you'd be way better off pulling that into a hashed file with a hard-coded key) or for doing something that requires no 'where clause' - for example, when selecting NEXTVAL from a Sequence object in Oracle from DUAL.