I heard back from Netezza last night and got a very reasonable response that I would like to share.
"..have not published a reference architecture because unlike other
RDBMS/Server/Storage combinations there are a very limited number of variables
or tunables, which effect Netezza performance and these are all covered in the
Netezza training course delivered with all installations and also available to
partners every month at Netezza HQ."
As we are working with a client who has Netezza as a major component of their architecture we are going to look into this further, and I appreciated a response. I think the disruptive innovation that the Appliance market is bringing is amazing and applaud it. Now I just need to understand where it fits – truely.
Your thoughts are the very reason I believe so strongly in an overall reference architecure to help bring all of these diverse possiblities into play while still leaving room for us to learn and evolve.
As you say as things evolve and develop we will be faced with things like appliances or new data integration tools that will challenge our previous roadmap decisions and the new tool developers will be excited and so will the end customers who win by being able to do things previously not possible.
I don’t think anyone can argue that Netezza is fast and inexpensive. The challenge most organizations will face in adopting an appliance strategy is the lack of maturity of the rest of the Netezza (or other appliance vendor offerings) business and technology products. I’d advise anyone considering choosing Netezza to do the due dilligence to ensure that they’ll receive the level of support their organization requires (commom sense with any vendor, really)
I suspect, more conservative organizations will be hesitant to adopt the Netezza model until they see a little more history, a stronger support organization and proven results.
I think that “disruptive innovation” is the best way to describe Netezza and its impact on the data operations of large organizations.
As a technology, the really interesting thing that Netezza does is it radically changes the processing power from the ETL server back to the database server. Kind of old school, DBA’s seem to love it. Things really change when it suddenly becomes much faster to do things with SQL (ELT) instead of pulling out the data and transforming it (ETL).
Of course, you then have issues around integration and information quality. I have seen a lot of operations that would take 10 hours to finish in ETL take 10 seconds to finish in Netezza. When you have that much more power then you can deal with these types of issues.
I think that ELT and SQL-ETL tool companies like Informatica and Sunopsis are loving this development, while flat file oriented companies like Ab Initio are going to have to adapt.
Netezza really changes the game for Informatin Architecture, something we discuss a lot on my own site http://architected.info/blog