Combining Transactional and Analytical Workloads with HW-Accelerated PostgreSQL in OVHcloud

Combining Transactional and Analytical Workloads with HW-Accelerated PostgreSQL in OVHcloud

• 2.
  PRESENTEDBY ENGINEERED ANALYTICS WITH POSTGRESQLIN OVHCLOUD JEAN-LOUIS QUEGUINER – OVH - TECHNICAL DIRECTOR BI/AI KARSTEN RÖNNER – CEO SWARM64 MARC GAUCHERON – INTEL ROOM 1 2.30 PM
• 3.
  OPEN-SOURCE IS THENEW WAY
• 4.
  RISE AND FALLOF CENTRALIZED DATA OFFICES
• 5.
  MMMM… I’VE SEENTHIS BEFORE, .. BUT WHERE ? CENTRALIZED : I’ll be managing your Machines DE-CENTRALIZED : I’ll open a dedicated department under IT control to help you manage your business. DISTRIBUTED : I’ll provide you the toolbox so that you are autonomous (e.g. a white labeled cloud) IT in companies Data in companies 90’s 00’s 10’s 00’s 10’s 20’s CENTRALIZED : All Data in one place. DE-CENTRALIZED : Provide BI resources at business level to help them leverage business insight. DISTRIBUTED : Provide tools to make business leverage data value.
• 6.
  WHAT ARCHITECTURE TOCONSIDER Extract Transform Load=Extract Load=Extract Load=Extract Load=Extract Central Data Office Company Level Decentralized Data Offices Business Unit Level Transform Transform Load back to BU Transform Load front office Explore Load back to Company
• 7.
  NOT INVENTED HERE
• 8.
  USE CASE: THE NIGHTWAS NOT ENOUGH Customer: Leading Financial Services Provider for Consumers Value add completely data driven: More Revenue, better Marketing, better Matching Swarm64 powers customer‘s entire Data Warehouse and Analytical Database Operational Database Extract TransformLoad Staging Tables for Nightly Run Production Data Warehouse Business Intelligence S64 DA for PSQL S64 DA for PSQL Query
• 9.
  DEMO VIDEO Input Sources Who wantsto wait 5 hours for the data ?
• 10.
  EXTENDING OPEN-SOURCE DB AnyStorage Architecture Application SQL Database Transactional Tables S64DA Tables
• 11.
  EFFICIENCY DEMANDS PERFORMANCE Specialized acceleratorsare a standard part of today’s IT infrastructure Efficient acceleration speeds up the entire data pipeline FPGA- Compression 3-5x FPGA Data Processing & Arrangement 1-5x Row-Column Self-Ordering Data-Structure 3-20x FPGA-Enabled Algorithmic Optimization FPGA-Enabled CPU Parallelism Data Sources 1x 1-3x 1-20x Data Storage / Data Ingest Pipeline
• 13.
  BUILD YOUR SOLUTION Network/ Direct- Attached Storage Load Balance Compute ProcessingIngest Input Stream Analysis JDBC/ ODBC Rest API Management and Monitoring
• 14.
  BENEFITS: EFFICIENCY, EASE, VERSATILITY Efficiency Consolidate infrastructure & workloads on same platform: transactional – analytical – streaming – real-time – key-value Ease  Acceleration Pipeline completely Invisible to Application Layer Versatility  Integrates Seamlessly into IT infrastructure
• 15.
  WANT TO KNOWMORE ? Pay a visit to Intel Booth to show solution live. Visit our OVH LAB and register for the SWARM64 trial