Indexima - IT Press Tour #59 Dec. 2024

Transcript

1. 2024 Speed-up your BI on Snowflake Zero Time to Data
2. Team Who are we ? Zero Time to Data
3. Nicolas KORCHIA CEO 15 years leading IT, Analytics & Big Data teams to excellence Sébastien Farges CTO 12 years leading projects within Accenture 5 years in Startup Emmanuel DUBOIS CRO 20 years in Corporate Account Selling with constant exceeded results 8 years in startups development A DATA EXPERTS TEAM
4. 15 customers 1.2M€ ARR 2020 2024 Next 2.0 Snowflake Version 2.0 DataBricks Version Inital problem 2015 2017 2018 2020 1.3M€ Seed 1.0 Hadoop Version 1.0 Saas + All DB From a company problem to a Saas Company
5. The BI Paradox Simple but complex ? Zero Time to Data
6. “65% of companies do not exploit their data.” IOT CIO inquiries “Data Analysts spent 48% of their time creating and updating reports versus 18% on communicating the results to the business.” Deloitte survey of 600 global leaders on management reporting “90% of data is trashed because of incapability to be exploited.” AWS Companies still struggle with their data…
7. From Data to Decision
8. COLLECT TRANSFORM ACTIVATE From Data to Decision
9. The BI Paradox Simple output Simplifying decision Human readable Internal complexity Intricate data modeling Complex queries Hard consequences Hard work in data enginering Data model always reworked Delay for « simple dashboards » Slowness of Dashboards
10. Pre aggregation For maximum perfs and minimum computing scale Zero Time to Data
12. SQL SQL SQL SQL “Select blablabla… Group By blablabla… ”
13. Whatever the size of the data, Human read Small Data Pre-aggregation : for sure !
14. Raw Data Aggregates Keys STORE results combinations of recurrent question To insure performances and save the planet Old Fashion - Cube - Extract - DataMart New wording - Materialized Views - Dynamic Tables Pre-aggregation : for sure !
15. Indexima Business Intelligence AI Assistant Zero Time to Data
16. Indexima is a unique data software that leverages ML and AI to solve the performance, cost, agility dilemma of analytics
17. CREATE Automatically Dynamic Tables OPTIMIZE Disk Scan ACCELERATE Response Time PREDICT Data Usage DYNAMIC TABLES MACHINE LEARNING Optimized in-db structures that accelerates the display of dashboards x1000 and reduces the scanned data Continuous comprehension of data usage to adapt and create the most accurate Dynamic Tables layer Indexima - main concepts
18. LIVE Demo NYC citibike Zero Time to Data
19. What are the options ? SQL SQL SQL SQL Direct access to raw data in Cloud DB 20s avg delay
20. What are the options ? SQL SQL SQL SQL Direct access to raw data in Cloud DB 20s avg delay Boost your snowflake WH Manualy modelize data
21. Let the AI speed up your BI SQL SQL SQL SQL At the beginning Indexima just acts as a proxy Direct access to raw data in Cloud DB 20s avg delay
22. SQL SQL SQL SQL Accelerated by Dynamic Tables After a few minutes Automatic creation of Dynamic Tables Let the AI speed up your BI <<1s average delay
23. SQL SQL SQL SQL Accelerated by Dynamic Tables Ultimately Indexima creates a perfect layer of Dynamic Tables. It speeds up the dashboards and allows not to sollicitate the raw data anymore. Let the AI speed up your BI <<1s average delay
24. PERF x100 Let the AI speed up your BI COST SAVED NO DEV
25. Open your board, change your url
26. do the magic for you Open your board, change your url Dynamic Table Analyser Query Routing
27. Products Indexima 1.0 and Indexima 2.0 Zero Time to Data
28. Indexima 1.0 BI expert Warehouse Since 2017 Analyzer Query Routing HyperIndexes Storage Indexima - store fined grained data - route queries - analyze queries to optimize - store aggregation layer in memory
29. Indexima 2.0 Snowflake Autopilot Starting 2024 Indexima - delegate fined grained data in DB - route queries - analyze queries to optimize - delegate aggregation layer in DB LIGHTER INFRASTRUCTURE LESS CHANGES FOR CUSTOMER NO DATA COPY – NO RIGHTS DELEGATION ADAPTED TO CLOUD DATA Analyzer Query Routing
30. 1.0 Automatic SQL optimizations + Aggregation layer copied in memory All databases 2.0 Automatic SQL optimizations + Aggregation layer in Database Only Snowflake The products
31. Use cases About customers real usage Zero Time to Data
32. Retail use case Max 144,2s Max 38,1s Max 11,5s Average 50,3s Average 12,4s Average 3,7s X-Small Medium X-Large X-Small Max 1,0s Cost by hour: 8$ Cost by hour: 32$ Cost by hour : 2$ INFRA costs minimal SPEED x100 faster + Average 0,2s DataViz : Tableau | Data Set : 1 Billion Rows | Infra : AWS Cost by hour: 2$
33. Finance Objective Analytic environment • Data Viz Tool: • Data Model: • Data volume: Key Results • Performance: • Snowflake cost: • DT Touch: E-commerce Transportation ++ X5 ISO 80% Star schema Star schema Denormalized ++ X10 -7% 90% ++ X100 ISO 100% Automation Cost/perf balance Performance Customers use-cases
34. On Prem Customers Saas platform customers 1.0 2.0 Customers
35. Go to market What about other softwares ? Zero Time to Data
36. Data warehouses Analytics Engine Dataviz tools Our competitive ecosystem BRUTE FORCE Need client to load data in house Linear correlation between size and price FEDERATION Focused on easing data access Not focused on performance and scalability COST OPTIM Focused on WH optimization Not focus on SQL Performance AGG LAYERS Full ingested Not automatic
37. Data engineer Team Analytics Team • Performance issue solver : time to insights • No back and forth with IT • Self-Service BI on all data sets • Ease of adoption • Data Cloud Cost reduction • Automatic Optimization • Automatic redirection • Ease of deployment They care about Sponsors – 2 different entry points
38. • Leads gen accelerator • Deal Influence • Service workforce Service Partners Technology Partners • Technical Certification • One clik connectivity • Customer confidence Strategy Partner GoToMarket - Partnership • Leads gen accelerator • Deal Influence • Marketplace
39. Pricing
40. Next Zero Time to Data
41. Product Roadmap Query Routing Materialization Analyzer Materialization in DataBricks Engine for Hadoop materialization Virtualization multi-sources Automatic WareHouse Tuning Materialization in DBT Analyzer Genetic Algorythms Cost / Perf control Panel monitor and forecast Semantic layer abstraction
42. Stop presuming what will be used, leave freedom to end user. AI for sustainability: automatically reducing byte scann for BI usage. Still the same 2 mantras since years
43. THANK YOU nico@indexima.com Zero Time to Data