Netezza is a not a transactional system. Intake for a Netezza machine is best served in bulk, at a gigabyte of data at-a-time or more. Several-records-in and several-records-out are a transactional model, where Netezza deals in bulk, millions and billions-at-a-time.

When you say you want to port your SQL statements, this is probably not completely true. You want to port your functionality to Netezza, but generally speaking by the time a system has

so run-its-course and is in need of performance boosts, several iterations of performance tuning have already been applied. And these are entirely artificial when moving over to Netezza.

Here's an example: If you take the standard cursor-based stored procedure, let's say you have a cursor pulling from a table and passing it through four business rules before putting it back. In one

transactional system using 3400 members of the cursor, it took 30 seconds per operation, translating into hours of time. In Netezza, we would take these rules and "stand them on their side", passing

all 3400 records through one rule, then the next and so forth. Each operation takes mere seconds, effectively deriving the same answer in less than a minute.

So in the end, you will have to think about what in your existing system is being processed "in bulk" and this is what you would push over to Netezza. Some of it, your transactional stuff, may remain

behind and require you to pump data between the two. That's okay, because that's how most reporting system work.

You can't do anything 1 row at a time on Netezza.  There are no indexes to allow you to get good performance.  At best you might get 10 queries per second.  Netezza is designed to process massive amounts of data in bulk.  You can process millions of rows per second, but your application has to be written to process sets of data at a time, not 1 row at a time.

Typically you would bulk load a big set of data into one or more staging tables, process the data using SQL in bulk operations, then load the results into your target tables.  It is actually far easier to write an application in this way than it is to do row at a time processing, but it is a very different mindset.  Netezza handles all the performance issues for you because every SQL statement always runs in parallel across 10s to 100s of cpus and disks.