Netezza Claimed 100x Faster Than Oracle Exadata

According to Oracle website:

The HP Oracle Database Machine is a complete system, including software, servers, networking and storage designed to run multi-terabyte data warehouses at least 10x faster than conventional data warehouse systems. At the heart of this system, is the HP Oracle Exadata Storage Server, which has smart storage software that offloads data-intensive query processing from database servers closer to the data. This results in much less data getting sent over fast Infiniband interconnects, dramatically improving both query performance and concurrency. The HP Oracle Database Machine runs Oracle Database 11g and Real Application Clusters on Oracle Enterprise Linux, and also includes complete Infiniband infrastructure. It's simple and fast to implement, very cost-effective, and can linearly scale storage capacity, processing power and network bandwidth as your data warehouse grows.

Oracle Exadata is a family of high performance storage software and hardware products that can improve data warehouse query performance by a factor of 10X or more. Oracle Exadata Storage is a combination of smart storage software from Oracle and industry-standard hardware from HP. Overcoming the limitations of conventional storage, Oracle Exadata uses a massively parallel architecture to dramatically increase data bandwidth between the database server and storage. In addition, smart storage software offloads data-intensive query processing from Oracle Database 11g Servers and does the query processing closer to the data. The result - faster, parallel data processing and less data movement through higher bandwidth connections.

The HP Oracle Exadata Storage Server is based on the HP ProLiant DL180 G5 server, and is a fast, reliable, high-capacity, industry-standard storage building block. With a choice of SAS or SATA drives and a storage capacity up to 12 TB per server, it has Oracle Exadata software pre-installed. In addition to extremely fast query processing for your large data warehouses, the massively parallel architecture offers linear scalability and mission-critical reliability.

Oracle Exadata Storage provides the foundation for building dynamic storage grids, and is the building block for the HP Oracle Database Machine. Designed for large, multi-terabyte data warehouses, the HP Oracle Database Machine is a complete package of software, servers, and storage. Simple and fast to implement, it has the power to tackle large-scale business intelligence problems immediately and can scale linearly as your data warehouse grows.

Following is challenge from Netezza.

The HP Oracle Database Machine is primitive, slow, expensive, closed, power-hungry, cannot handle advanced analytics and requires significant management overhead.

But other than that, it’s not a bad start for Oracle!

For over five years, Oracle users have been sharing with Netezza their concerns about the performance of Oracle as either an enterprise data warehouse or for specific analytic applications. And as data volumes grow, analytics become more complex, users demand faster response and cost reduction initiatives become rampant, Oracle users have simply been unable to keep up. So much so, in fact that over 100 of these users turned to Netezza and our data warehouse appliances for a solution.

The new Database Machine is Oracle’s response to Netezza. Like Netezza did some seven years before, Oracle has now realized that an architecture where SQL processing is carried out close to the disk yields all sorts of performance gains. Data not required in a query is filtered out at the source and is prevented from causing a bottleneck between storage and the database where the complex processing is carried out. And just like Netezza did, Oracle has also recognized the benefits of massively parallel processing.

But it’s here that the similarities end and fact can be confused with fiction. With the HP Oracle Database Machine, what are you really getting? Users will find that the Oracle world hasn’t changed that much in the last five years.

In writing, Oracle claims a 10X or more performance improvement over its previous systems. Netezza already delivers up to 100x or more performance over conventional systems, of which Oracle is a prime offender. Companies such as Ahold, Epsilon, iBasis, NYSE Euronext, The Carphone Warehouse and Virgin Media have already realized the benefits.

Oracle’s one-to-many configuration of processor to disk - the Exadata nodes are organized with eight processor cores “sharing” 12 disk drives - is less efficient and with just 200 processors so they aren’t able to take advantage of parallelism to the same extent as Netezza. Netezza’s game-changing performance capability comes from our one-to-one mapping of processor to disk and the fact that we are massively parallel (nearly 1000 processors in our largest machine). Netezza has also put much more sophisticated processing capabilities into each of those 1,000 processors, or more suitably referred to as “intelligent storage nodes.”

As a first generation machine, Exadata is primitive and is limited to carrying out simple SQL projection operations within the storage. The bulk of processing is still carried out in the database itself and, as a consequence, costly amounts of raw, unprocessed information are sent to the SMP host processors in the Database Machine for further processing. By comparison, the Netezza Performance Server® (NPS®) system typically performs well over 95% of all of the work required to complete a given query response within the massively parallel array of 1,000 intelligent storage nodes, adding to the performance benefit.

All of the above leads to higher performance in the NPS system. While Netezza makes claims of 100x performance gains, it is not uncommon to see performance differences as large as 200x to even 400x or more when compared to existing Oracle systems. Oracle’s 10x performance claim is weak in comparison.

Oracle processes SQL queries. With Netezza’s OnStream™ capabilities, we have "opened up" our patented streaming architecture so that developers can exploit the same massively parallel processing that we do for high-performance, non-SQL, advanced analytics. The NPS system is fully extensible, and complex user-defined functions can be executed within Netezza’s massively parallel storage array. A sophisticated programmer’s workbench allows functions to be written quickly and easily with the massively parallel considerations being handled automatically by Netezza. Netezza customers all over the world are benefitting from this, which results in game-changing breakthroughs in performance even over an already high-performing vanilla Netezza implementation. Today, there are over 100 partners who have written advanced analytics functions for risk simulation, dynamic re-pricing for telecommunications, customer scoring and complex transformational logic, fuzzy name matching for intelligence and fraud detection and others, that take advantage of this capability.

Netezza is supplied as a single system ready-to-go in less than two days. This has a significant impact on time to value (TTV) which is important in today’s economy. In contrast, the Database Machine is supplied by two different companies, dispatched piecemeal on separate days. When all the parts arrive on site, you can request an engineer to come and set it all up. As a piecemeal solution, implementation is no different to that of a regular Oracle set-up: time consuming, expensive, inefficient and painfully slow.

Because Netezza’s ANSII standard SQL RDBMS was designed to run Netezza’s unique hardware architecture, it does not need the complex performance tuning mechanisms that you find in Oracle. This results in administrators of the system and the database having to deal with and maintain, literally, thousands fewer objects in the system. As a consequence, a key aspect of Netezza is simplicity. Netezza systems are pre-installed, pre-built, pre-configured, require minimal administration and run orders of magnitude faster than Oracle.

Will Oracle takeover Sun Microsystem?

IBM is no longer interested in buying smaller rival Sun Microsystems Inc at any price, due to concerns that such a deal would draw intense regulatory scrutiny. If Oracle seriously want to build their own Database Machine, they can afford to takeover Sun Microsystem.