O slideshow foi denunciado.
Utilizamos seu perfil e dados de atividades no LinkedIn para personalizar e exibir anúncios mais relevantes. Altere suas preferências de anúncios quando desejar.

The Benefits of Fast Business Intelligence // Amir Orad, Sisense [FirstMark's Data Driven]

682 visualizações

Publicada em

Amir Orad, CEO of Sisense, presented at FirstMark's Data Driven NYC on January 19, 2016. Orad discussed Sisense's approach to speed in business analytics.

Sisense creates business analytics software that lets you easily prepare, analyze and visualize complex datasets using one Single-Stack solution.

Data Driven NYC is a monthly event covering Big Data and data-driven products and startups, hosted by Matt Turck, partner at FirstMark.

FirstMark is an early stage venture capital firm based in New York City. Find out more about Data Driven NYC at http://datadrivennyc.com and FirstMark Capital at http://firstmarkcap.com.

Publicada em: Tecnologia

The Benefits of Fast Business Intelligence // Amir Orad, Sisense [FirstMark's Data Driven]

  1. 1. Big Data Biz Analytics Innovation & Disruption Amir Orad Sisense CEO Feb 2016
  2. 2. Quick Background Inventor/co-founder of Risk Based Authentication company Cyota (RSA) CEO of $200M Financial Crime analytics company Actimize (NICE) Leading Sisense to simplify Complex Data Analytics Columbia University MBA
  3. 3. What Do Five Data Geek Students Dream About
  4. 4. BEERS & CHIPS
  5. 5. Technology Disruption Traditional BI 1995 In-Memory BI 2005 In-Chip BI 2015
  6. 6. Memory Bandwidth – Data Size vs Speed Too SlowX50 X10 L1 cache L2 / L3 cache RAM DiskDistance from L1 = slowdown
  7. 7. Data Beer IN ORDER TO UNDERSTAND IN-CHIP ANALYTICS LET’S ASSUME THAT:
  8. 8. Memory bandwidth 2 L1 cache Home fridge Distance Immediate Customer x1 L2 / l3 cache Shop Distance Bicycle Customer x10 Ram Supermarket Distance Car Customer x50 Disk Brewery Distance Airplane Customer If data equals beer then data storage units equals
  9. 9. Vectorization & Cache Awareness L1Cache FirstintoRAM OP 100 4K (Values) 100 4K (Values) 100 4K (Values) Result Vector Push Back To RAM 100 4K (Values) SIMD REGISTER Apply Operation On 4/8 Data Elements Simultaneously OP OP Column4 100 4K (Values) ResultVector 100 4K (Values) Column1 100 4K (Values) Column2 100 4K (Values) Column3 100 4K (Values)
  10. 10. Our Technology In memory columnar execution mode CACHE aware query kernel CACHE aware decompression Instruction recycling & learning algorithms LLVM based compiler with SIMD support Full 64BIT support Columnar storage
  11. 11. SPEED! STRATA AWARD Analyzing 10TB of data In 10 seconds On a single node on a standard Dell Server
  12. 12. Are We Solving the Real Problem?
  13. 13. Breaking an Assembly Line Tradition
  14. 14. Need DBA to build database Define what data will be queried Join tables upfront Normalize and create a star schema Why?
  15. 15. Surprising Benefits Handle complex data faster, cheaper, easier Boost performance 10X-100X; Cut HW reqs Eliminate & simplify data preparations NO DBA, Join, Index, Star Schema Fast to deploy; Agile to change Self service for everyone - Biz & IT Shrink TCO & time to insight
  16. 16. Technology Disruption Results DW, OLAP Complex “Expensive” mash-up TB Scale Months Traditional BI 1995 In-Memory Simple for Biz Manual mash-up GB Scale Weeks In-Memory BI 2005 In-Chip Simpler for Biz & IT Ad-hoc mash-up TB Scale Days In-Chip BI 2015
  17. 17. A Dream Comes True – 1000+ Clients
  18. 18. Recent Enhancement - R Most time wasted on data prep Democratize advance analytics
  19. 19. Lessons Learned Dream BIG Refine, refine, refine benefits Don’t automate, obliterate! Disrupt, don’t improve
  20. 20. Thank You Try It Out Free Trial www.sisense.com
  21. 21. Thank you

×