How can your business unlock the enormous value stored in the mountains of data you possess?
This is the question the IBM partnership with Fuzzy Logix seeks to answer.
Without a doubt it all begins with a robust hardware platform, the ability to leverage quantitative models to analyze large sets of data, and a way to present the results in an easy to understand and actionable format. By marrying the Netezza (PureData System) platform with Fuzzy Logix’ quantitative models and web-based solutions, together we provide turn-key solutions capable of handling infinitely large amounts of data while eliminating the need for long, capital intensive development projects. Problem solved. Opportunity realized.
For companies adopting the Netezza platform who have existing quantitative models, Fuzzy Logix also provides consultative services that leverage our vast array of technical and modeling experience to help you optimize your existing models and add new capabilities. These services leverage our technology while also making the most of our human capital in support of your existing team.
- Logistic Regression – 100,000 Runs with 15 Random Variables Each – Netezza analytics 50 minutes; Fuzzy Logix on the same Netezza machine with the same data 33 seconds.
- Principal Component Analysis – Netezza analytics 4 minutes; Fuzzy Logix on the same Netezza machine with the same data 6 seconds.
- Kmeans- Netezza analytics 3 minutes 44 seconds; Fuzzy Logix on the same Netezza machine with the same data 12 seconds.
- Market basket analysis with 80,000 products – SAS 20 hours; Fuzzy Logix 2 hours
- Healthcare provider scoring – SAS/Oracle 6 weeks, 25 jobs; Fuzzy Logix 1 job in 7 minutes.
- Preventative healthcare – 4,500 variables, 3 million + rows of data. Large regression. Previously not possible to run – executed in 3 minutes with our models
- Banking – value at risk for equity options – simulation – 2.5 billion simulations in less than 3 minutes.
- Portfolio Management – IRR for 15 million portfolios – Traditional solution 12-15 hours; Fuzzy Logix 7 minutes.
- Credit Card Risk Modeling – Regression with 2,000 variables 5X improvement. Tested up to 11,500 variables. R 990 seconds, Fuzzy Logix 172 seconds.
- Customer Scoring – 10 hours with SAS; 10 minutes with Fuzzy Logix.
DB LYTIX™ FOR NETEZZA OR IBM PUREDATA SYSTEMS
Massively parallel architecture and access to a bit of “secret sauce” allow DB Lytix™ to run amazingly fast on the Netezza platform. The reason for this incredible performance can be understood after exploring a bit of the Netezza lexicon.
Netezza’s AMPP™(Asymmetric Massively Parallel Processing) architecture marries a SMP (Symmetric Multiprocessing) host with MPP (Massive Parallel Processing). As programs are executed, the host compiles the code and distributes queries across individual SPUs (Snippet Processing Units) to run in parallel. Once distributed, the SPUs perform the processing. When the processing is complete, the SMP host aggregates the results. The SPU is a complete computer with a processor, disk storage, memory and a key piece of the “secret sauce” of Netezza. By using the combination of SMP Host and SPUs, parallelized SQL queries execute with blazing speed.
“SECRET SAUCE” – THE FPGA
Another acronym introduced by Netezza to the data warehouse world is FPGA (Field Programmable Gate Array). FPGAs are semiconductor chips that can be programmed and have been used for a wide variety of applications for many years. Netezza is the first company to leverage their power to process streaming data in a data warehouse appliance. In traditional systems, all the data for a query is moved and then the “where” clause is processed. With Netezza, instead of moving a huge set of data, the FPGA processes the “where” clause as data streams off of the disk, so only the data needed for processing is moved to the next step.
To access the power of the architecture, Netezza allows users to write UDFs (User Defined Functions) and UDAs (User Defined Aggregates) – both known as UDxs. UDA’s return one value for multiple rows, whereas UDF’s return one value per row. UDA’s and UDF’s can be accessed through simple SELECT statement and UDF’s can be embedded within UDA’s and UDF’s. Also, it is possible to embed UDA’s within UDF’s as arguments and vice versa.
With UDxs, Netezza allows you to write processing logic, such as data mining algorithms, to the SPUs. Netezza calls this OnStream™processing. By implementing this and other technologies, Netezza achieves speeds which are 10-100 times faster than traditional solutions.
While it is possible to create your own UDxs, Fuzzy Logix DB Lytix™ software package contains hundreds of prebuilt math, statastic, data mining and financial models.
(AMPP™+ SPUs + FPGA + UDx) + DB Lytix™ = STUNNING PERFORMANCE
By combining the Asymmetric MPP™architecture with SPUs and FPGAs, and then giving access to the SPUs via UDAs and UDFs, Netezza gives users a strong platform to achieve very high processing performance.
Pairing the platform with the DB Lytix™ library leverages the architecture of Netezza to its fullest extent and delivers results with stunning speed and accuracy.
Please contact us to discuss how our solutions improve business performance.