Timothy raises an excellent point -
For data warehousing, I hope we're not comparing a purpose-built, massively parallel environment to a general-purpose SMP-based RDBMS like Oracle, because the traditional RDBMS cannot hope to keep up with data warehousing, with or without Netezza in play.
What are Netezza's weaknesses for data warehousing? None - in fact simple data layouts and elimination of structural engineering (for performance) are significant strengths
What are Netezza's weaknesses for transactional processing? Considering that it is not built for transactions (at all), then this isn't a weakness, it's missing by design.
If I were to look at the marketplace for the common SMP-based RDBMS, I would find:
Oracle and SQLServer:
for transactional processing - sweet spot, great products
for data warehousing - second rate (and always has been) - even Business Objects calls them secondhand technologies for data warehousing. They require too much structural engineering to get the data to perform even at marginal speeds for both loading and querying. They were (and are) industry placeholder-technologies for those who are waiting for something better. It's here - It's Netezza.
For data warehouse scaling - SQLServer breaks down to intolerable lethargy at about 50 million records, about half a terabyte. Oracle can go as high as 2 terabytes and 200 million rows before it starts to weeble/wobble without falling down, still exhibiting intolerable lethargy. Supporting anything larger requires millions of dollars in additional hardware.
In Netezza, billions-of-rows, multi-terabyte warehouses are the sweet spot. The RDBMS is in the rear-view-mirror and shrinking rapidly
