Technical Overview Ernesto Herrera Vectornova CTO July 2010 Copyright Vectornova 2004-2010 / All rights reserved
Extreme Performance   data engine increasing   performance   with   power empowering   management   with   performance
Vectornova  is the  performance leader  among the new generation of high-speed  Columnar Data Engines. Our  VECTOR BASED  application enable organizations to analyze terabytes of information in seconds.
vectorSTAR® ,  vectorized relational database , eliminates the problems of slow response times, low performance, vast complexity, and huge cost associated with the creation and operation of  very large data warehouses , data marts and  multidimensional data cubes.
vectorSTAR ®  extreme speed and performance  –often  1000x better – with maximum security.  Its powerful  V-SQL  plus  J & R software  based analytical, statistical and mathematical capabilities, breakthrough architecture plus a simple user interface make it the smart choice while working in almost any hardware  saving thousands of dollars .
vectorSTAR ®  is not only the smart but the ultimate choice both as a  high-performance data-mart  complementing existing infrastructure, or as the core of a truly  enterprise-wide data warehouse.
vectorSTAR ®  will transform your system in a stay-of-the-art high-performance  super computer  with minimum investment and maximizing ROI.
Vectornova  has offices in Canada, Austria and Mexico plus distribution in U.S.A., Germany,   Singapore, Hong-Kong, France, England, Spain, Brazil, Colombia, Argentina and Chile..
vectorSTAR ®  has a joint technology partnership with  J Software  (Canada)  Our  R&D team  include 35 brilliant scientists,  engineers, designers, business analysts,  mathematician & statistician, plus dozens of collaborators.
vectorSTAR ®  Facts Product research started in 2001. Building Application started in 2003. First Customer deployment in 2004: State Police Intelligence System in 23 counties. Embedded with OEM facial-recognition application. Lowest downtime recorded for a server: 0.01%. Largest cardinality deployment: 3.5 billion records, scaling up to 15 billion. Largest storage size deployment up to date: 32 Terabytes. Largest running daily mission continuous critical operation: RZB (Austria, 2007)
vectorSTAR ®  Deployed Applications: Government: Law Enforcement, Internal Revenue Service. Banks: VaR (Value at Risk), Credit Card data mining. Telecom: CDRs (call detail records) Large Retailers: Basket Analysis, Line-item level data mining and reporting, Operational BI. Manufacturers: Operational BI. Insurance: Historical detail records, Multi-variable sinister reports, Operational BI.
Data Management Environment vectorSTAR Vertica Sybase IQ BrightHouse Kognitio Paraccel KBD (1010data) Teradata Netezza Datallegro Greenplum HP Neoview  Oracle MS SQL Server Sybase ASE IBM DB2 Informix MySQL PostgreSQL Essbase Oracle Express Cubes Google’s Bigtable Key-Value Red Brick MS ROLAP ObjectStore  Versant Caché db4obj Object Oriented IMS Cullinet Network  Hierarchical OLTP Appliances Columnar ROLAP Relational Model Non Relational DBase III Access XBase Spreadsheet Lotus 1-2-3 Excel
vectorSTAR ®  is a unique VECTOR Columnar Relational Database. Speed:  100x, 500x and up to 1,000x faster than traditional RDBMS.
vectorSTAR ®  cost of ownership allows implementations at half and up to 1/10 of any other application. Price/Performance:  1,000x better
vectorSTAR ®  has great flexibility: interactive SQL, sophisticated data types, zero passive footprint, web access, SAN. Simplicity:  our technology leaps make things easier and developing or integrating applications is fast and natural.
vectorSTAR ®   Architecture Full   Columnar Model Array-Based Column Storage , vs. Set-based storage, thus avoiding the need for indexes to correlate them Memory-Mapped File I/O  vs. Buffered File I/O (this is state of the art in HPC techniques and greatly increases performance and reliability) Multidimensional Vector Programming/Querying  language that facilitates the natural application of parallel operations on huge arrays of data
vectorSTAR ®   Usage Scenarios
vectorSTAR ®   at its best high cardinality  +  light-record  databases  billions of records medium-cardinality   +  heavy-record  databases dozens of terabytes, millions of documents  and multimedia
vectorSTAR ®   Highlights Designed from the ground-up as a  fully 64-bit  application for the  x86-64  architecture (both Intel and AMD) Supports a close equivalent of the complete  ANSI SQL2003  and  ANSI SQL2008  functionality via  Vector SQL Plus a significant subset of ANSI SQL92 via ANSI SQL APIs for  HTTP, HTTPS, Java, .NET (C# & VisualBasic), C, and C++  programming languages Real-time, high-performance, easy-to-use, bi-directional link with  Microsoft   Excel
vectorSTAR ®   Features Supports large variety of  data types   ( biometrics, images, video  and  sound ) - XML, PDF, IMAGES & VIDEO Format - FINGERPRINT  & FACEPRINT. Supports  read-only   tables – protect sensitive data Supports  encrypted  tables – further protect  sensitive data Extensive  platform support Linux  (Suse, Redhat, Ubuntu),  Windows  (XP64, Server2003, Server2008)  Unix  (FreeBSD, Mac OS/X) Supports  32-bit  platforms for clients, small office servers, and  mobile  deployments ( Android  support by the end of the year) Hot backups
vectorSTAR ®   uniqueness Add/Remove columns at run time without stopping the engine Removing a column affects only apps using it Incremental renaming of columns Incremental SQL Shared result sets Biometric query integration Remote Query via UDP over Radio Frequency: l aptops and handhelds on police cars connect to  data warehouse using their comm radio and PDA systems
vectorSTAR ®   scalability Supersymmetric scales out with a rep/perf factor > 1 Superparallel   I/O column parallel No specialized tuning  is required to achieve high performance  It is  simple  to deploy and use effectively
vectorSTAR ®   speed means What does  1000x faster  represents? 0.8 seconds vs 13 minutes 8 seconds vs 2 hours and half 3 hours vs 125 days!  (that’s 4 months!) 18 hours vs 2 years!
 
 
 
 
 
Extreme Simplicity
vectorSTAR ®   hardware requirement Minimum Acceptable Node Dual Xeon/Opteron Dual Core 16 GB RAM 4 x 250 GB SATAII 10K RPM HDD Windows Server2008 Linux Red Hat Cost: < $2,500 USD
vectorSTAR ®   hardware requirement Entry Level Node Dual Xeon/Opteron Quad Core 32 GB RAM 8 x 250 GB SATAII 10K RPM HDD Windows Server2008 Linux Red Hat Cost: < $5,000 USD
vectorSTAR ®   hardware requirement Medium Level Node Dual Xeon/Opteron Quad Core 48 GB RAM 1 Areca 1280 HBA Controller Card 4 x 250 GB SATAII 10K RPM HDD 4 x 80 GB SSD  Linux Red Hat Cost: < $10,000 USD
vectorSTAR ®   hardware requirement High Level Node Dual Xeon/Opteron Six Core 64 GB RAM 1 Areca 1680 HBA Controller Card 16 x 250 GB SATAII 10K RPM HDD 4 x 80 GB SSD  FreeBSD Cost: < $15,000 USD
Demo show what can be achieved with a solid  quad-core 64bit  pizza box with 32GB RAM (no more than US$4,000 if you shop carefully) POC - Proof of concept takes  two-three weeks  for simple POCs, about 2-3 months for complex ones free of charge a mirror of  something that you already   do  but find   problematic  preferably with dozens or  hundreds of millions   of records and complex analytical joins, and back end calculations compare  the results: cost + performance + ease of development Pilot Project vector-STAR   data-mart paralleled  with current infrastructure Eased into production –  eliminates bottlenecks and further reduces risk

vectorStar-2010

  • 1.
    Technical Overview ErnestoHerrera Vectornova CTO July 2010 Copyright Vectornova 2004-2010 / All rights reserved
  • 2.
    Extreme Performance data engine increasing performance with power empowering management with performance
  • 3.
    Vectornova isthe performance leader among the new generation of high-speed Columnar Data Engines. Our VECTOR BASED application enable organizations to analyze terabytes of information in seconds.
  • 4.
    vectorSTAR® , vectorized relational database , eliminates the problems of slow response times, low performance, vast complexity, and huge cost associated with the creation and operation of very large data warehouses , data marts and multidimensional data cubes.
  • 5.
    vectorSTAR ® extreme speed and performance –often 1000x better – with maximum security. Its powerful V-SQL plus J & R software based analytical, statistical and mathematical capabilities, breakthrough architecture plus a simple user interface make it the smart choice while working in almost any hardware saving thousands of dollars .
  • 6.
    vectorSTAR ® is not only the smart but the ultimate choice both as a high-performance data-mart complementing existing infrastructure, or as the core of a truly enterprise-wide data warehouse.
  • 7.
    vectorSTAR ® will transform your system in a stay-of-the-art high-performance super computer with minimum investment and maximizing ROI.
  • 8.
    Vectornova hasoffices in Canada, Austria and Mexico plus distribution in U.S.A., Germany, Singapore, Hong-Kong, France, England, Spain, Brazil, Colombia, Argentina and Chile..
  • 9.
    vectorSTAR ® has a joint technology partnership with J Software (Canada) Our R&D team include 35 brilliant scientists, engineers, designers, business analysts, mathematician & statistician, plus dozens of collaborators.
  • 10.
    vectorSTAR ® Facts Product research started in 2001. Building Application started in 2003. First Customer deployment in 2004: State Police Intelligence System in 23 counties. Embedded with OEM facial-recognition application. Lowest downtime recorded for a server: 0.01%. Largest cardinality deployment: 3.5 billion records, scaling up to 15 billion. Largest storage size deployment up to date: 32 Terabytes. Largest running daily mission continuous critical operation: RZB (Austria, 2007)
  • 11.
    vectorSTAR ® Deployed Applications: Government: Law Enforcement, Internal Revenue Service. Banks: VaR (Value at Risk), Credit Card data mining. Telecom: CDRs (call detail records) Large Retailers: Basket Analysis, Line-item level data mining and reporting, Operational BI. Manufacturers: Operational BI. Insurance: Historical detail records, Multi-variable sinister reports, Operational BI.
  • 12.
    Data Management EnvironmentvectorSTAR Vertica Sybase IQ BrightHouse Kognitio Paraccel KBD (1010data) Teradata Netezza Datallegro Greenplum HP Neoview Oracle MS SQL Server Sybase ASE IBM DB2 Informix MySQL PostgreSQL Essbase Oracle Express Cubes Google’s Bigtable Key-Value Red Brick MS ROLAP ObjectStore Versant Caché db4obj Object Oriented IMS Cullinet Network Hierarchical OLTP Appliances Columnar ROLAP Relational Model Non Relational DBase III Access XBase Spreadsheet Lotus 1-2-3 Excel
  • 13.
    vectorSTAR ® is a unique VECTOR Columnar Relational Database. Speed: 100x, 500x and up to 1,000x faster than traditional RDBMS.
  • 14.
    vectorSTAR ® cost of ownership allows implementations at half and up to 1/10 of any other application. Price/Performance: 1,000x better
  • 15.
    vectorSTAR ® has great flexibility: interactive SQL, sophisticated data types, zero passive footprint, web access, SAN. Simplicity: our technology leaps make things easier and developing or integrating applications is fast and natural.
  • 16.
    vectorSTAR ® Architecture Full Columnar Model Array-Based Column Storage , vs. Set-based storage, thus avoiding the need for indexes to correlate them Memory-Mapped File I/O vs. Buffered File I/O (this is state of the art in HPC techniques and greatly increases performance and reliability) Multidimensional Vector Programming/Querying language that facilitates the natural application of parallel operations on huge arrays of data
  • 17.
    vectorSTAR ® Usage Scenarios
  • 18.
    vectorSTAR ® at its best high cardinality + light-record databases billions of records medium-cardinality + heavy-record databases dozens of terabytes, millions of documents and multimedia
  • 19.
    vectorSTAR ® Highlights Designed from the ground-up as a fully 64-bit application for the x86-64 architecture (both Intel and AMD) Supports a close equivalent of the complete ANSI SQL2003 and ANSI SQL2008 functionality via Vector SQL Plus a significant subset of ANSI SQL92 via ANSI SQL APIs for HTTP, HTTPS, Java, .NET (C# & VisualBasic), C, and C++ programming languages Real-time, high-performance, easy-to-use, bi-directional link with Microsoft Excel
  • 20.
    vectorSTAR ® Features Supports large variety of data types ( biometrics, images, video and sound ) - XML, PDF, IMAGES & VIDEO Format - FINGERPRINT & FACEPRINT. Supports read-only tables – protect sensitive data Supports encrypted tables – further protect sensitive data Extensive platform support Linux (Suse, Redhat, Ubuntu), Windows (XP64, Server2003, Server2008) Unix (FreeBSD, Mac OS/X) Supports 32-bit platforms for clients, small office servers, and mobile deployments ( Android support by the end of the year) Hot backups
  • 21.
    vectorSTAR ® uniqueness Add/Remove columns at run time without stopping the engine Removing a column affects only apps using it Incremental renaming of columns Incremental SQL Shared result sets Biometric query integration Remote Query via UDP over Radio Frequency: l aptops and handhelds on police cars connect to data warehouse using their comm radio and PDA systems
  • 22.
    vectorSTAR ® scalability Supersymmetric scales out with a rep/perf factor > 1 Superparallel I/O column parallel No specialized tuning is required to achieve high performance It is simple to deploy and use effectively
  • 23.
    vectorSTAR ® speed means What does 1000x faster represents? 0.8 seconds vs 13 minutes 8 seconds vs 2 hours and half 3 hours vs 125 days! (that’s 4 months!) 18 hours vs 2 years!
  • 24.
  • 25.
  • 26.
  • 27.
  • 28.
  • 29.
  • 30.
    vectorSTAR ® hardware requirement Minimum Acceptable Node Dual Xeon/Opteron Dual Core 16 GB RAM 4 x 250 GB SATAII 10K RPM HDD Windows Server2008 Linux Red Hat Cost: < $2,500 USD
  • 31.
    vectorSTAR ® hardware requirement Entry Level Node Dual Xeon/Opteron Quad Core 32 GB RAM 8 x 250 GB SATAII 10K RPM HDD Windows Server2008 Linux Red Hat Cost: < $5,000 USD
  • 32.
    vectorSTAR ® hardware requirement Medium Level Node Dual Xeon/Opteron Quad Core 48 GB RAM 1 Areca 1280 HBA Controller Card 4 x 250 GB SATAII 10K RPM HDD 4 x 80 GB SSD Linux Red Hat Cost: < $10,000 USD
  • 33.
    vectorSTAR ® hardware requirement High Level Node Dual Xeon/Opteron Six Core 64 GB RAM 1 Areca 1680 HBA Controller Card 16 x 250 GB SATAII 10K RPM HDD 4 x 80 GB SSD FreeBSD Cost: < $15,000 USD
  • 34.
    Demo show whatcan be achieved with a solid quad-core 64bit pizza box with 32GB RAM (no more than US$4,000 if you shop carefully) POC - Proof of concept takes two-three weeks for simple POCs, about 2-3 months for complex ones free of charge a mirror of something that you already do but find problematic preferably with dozens or hundreds of millions of records and complex analytical joins, and back end calculations compare the results: cost + performance + ease of development Pilot Project vector-STAR data-mart paralleled with current infrastructure Eased into production – eliminates bottlenecks and further reduces risk