Re: NETEZZA, A Data-warehouse tool

From: Dieter Noeth <dnoeth_at_gmx.de>
Date: Mon, 01 Jan 2007 22:15:18 +0100
Message-ID: <enbtjp$2or$1@online.de>

pankaj_wolfhunter_at_yahoo.co.in wrote:

> My management has told me to look into NETEZZA, a
> data-warehouse, tool.

...
> I tried on google but couldnt find much help.

You'll hardly find any independant information regarding Netezza, as everything (even the manuals) is covered by a non-disclosure-agreement.

> "Netezza Corporation produces a data warehouse appliance, Netezza
> Performance Server (NPS). Netezza claims that their product delivers
> 10-100 times the performance at half the cost of traditional data
> warehouse systems. The product competes with products from Oracle
> Corporation, IBM, Teradata and others"
>
> Is it really the case?

It's from the marketing dept, so it must be the truth ;-)

Preliminary note:
I'm working with Teradata, so my opinion is probably a bit biased :-)

For me Netezza looks like a clone of a Teradata V1, which was available in 1984:
Lots of independant parallel Processing Units (Netezza:SPU, Teradata:AMP) with dedicated disks (Netezza: 1, Teradata V1: 1 to several) using a Shared Nothing approach.

Netezza's SPUs run a heavily modified version of PostgreSQL on a proprietary OS, only the "host nodes" (Teradata V1: PE/Parsing Engine) run Linux.

Some issues to think about:

- data is not RAID protected -> if you loose one disk...
- cheap PC-class SATA disks instead of high performance SCSI
- no kind of index, everything results in Full Table Scans
- fast for aggregating columns large amounts of data from a single 

table, but slow for complex joins
- AFAIK no Referential Integrity, no Stored Procedures, no Triggers, no UDFs, no Materialized Views
- the company is just a few years old with less than 100 customers

Dieter Received on Mon Jan 01 2007 - 15:15:18 CST