Features - Enterprise Data Insights:

NETEZZA TAKES ON TERADATA -- GET IT RIGHT!
by William Fellows for the451.com

Netezza's challenge to Teradata is on the market. The company has begun shipping its Performance Server 8000 Series, a high-end massively parallel processing data warehouse engine designed for handling very large business intelligence requirements. Netezza says Teradata has had the space to itself for far too long and still hasn't got it right. The NPS 8000 server, it claims, can access more data, more quickly and at a fraction of the cost of a Teradata system.

Impact assessment:

The message

Teradata and IBM have had the large-scale data warehousing and BI market to themselves for too long. From scratch, Netezza has created a system designed to do what they do but acting on more data, faster and at a fraction of the price.

Competitive landscape

The chief competitors are Teradata and IBM. Netezza will have to work hard to gain any ground on either. Teradata is used by the world's largest data warehouse customers, such as Wal-Mart.

The451 assessment

Netezza may be cheaper and faster, but the best don't necessarily succeed, especially when the competition has been at it for years and the spending climate is as it is today. Who spends $1m with a startup without a track record? (And at $1m a pop -– it's not exactly a volume market.) The company will need customer proof points and at least two years of existence under its belt before it's going to get looked at by any potential customers.

Technology BI applications by vendors such as Business Objects, Cognos, MicroStrategy, SAS, Unica and others help companies gather information from their data through customer segmentation, Web log analysis, reporting, data mining and other analytical activities. BI tools such as Informatica and Ascential extract, transform and load data from data sources so they can be used for analyses. Backup and recovery tools such as Tivoli help provide reliability to database systems. The NPS server includes its own relational database, which these applications use to create and populate data warehouses. The NPS system uses common interfaces and standards (ODBC, JDBC), so customers don't have to restructure or reformat their existing data according to Netezza.

Netezza claims two key technical innovations. First, its core IP, the company's Intelligent Query Streaming design, optimizes the flow of information by placing up to 114 field-programmable gate-arrays close to the storage device, processing records as they come off disk and bringing over only the relevant information for each query. The company says this class of processing would usually overwhelm a general-purpose server.

Second, NPS is built using an asymmetric massively parallel-processing (AMPP) architecture. It's a hybrid SMP/MPP design. At the front end, an SMP-based host compiles queries into parallel execution plans and provides the processing power to sort, and aggregate, large sets of query results. On the back end, data is distributed across many nodes (or 'spindles') to minimize I/O latency and increase scalability. Query functions are implemented in silicon to optimize throughput.

Marketing And Finance

Netezza claims NPS delivers 10-20 times the performance for large, complex and constantly growing BI efforts, at half the cost of existing systems.

It says a Fortune 500 financial services institution needed to analyze a database with a massive 287 million rows for customer payment behavior and credit-worthiness to predict delinquencies. It took more than 40 minutes to run this query on that company's existing equipment. The NPS system ran the query in under a minute -– although it wouldn't name the company or incumbent supplier.

Netezza has raised $28.5m since December 2000 and has 70 staff. Its plan is to achieve breakeven by the end of next year, but it will need more investment before then. The company claims to have some revenue and a handful of customers that it hopes to be able to reference soon in order to build momentum.

Courtesy the451.com