building an iot kafka pipeline in under 5 minutes
TRANSCRIPT
Today’s Talk
IoT is big
Ka#a is popular
Real-&me pipelines need opera&onal data warehouses
How can we do this simply (in under 5 minutes)
MemSQL 3
Real Time and Opera.onalGo Hand In Hand
• Live applica+ons and embedded analy+cs
• Con+nuous data processing (not batch)
• Real-+me AND historical data together
MemSQL 5
Gartner: An Opera,onal Data Warehouse• Manages structured data
• Loads con0nuously for embedded analy0cs in applica0ons
• Supports real-&me data warehousing
• Func0ons as an opera0onal data store
MemSQL 6
Gartner: An Opera,onal Data Warehouse• Manages structured data
• Loads con0nuously for embedded analy0cs in applica0ons
• Supports real-&me data warehousing
• Func0ons as an opera0onal data store
• Query op&miza&on plays a role (new)
• Many queries are repe00ve, mul0plying effects of op0miza0on
MemSQL 7
Magic Quadrant for Data Management Solu4ons for Analy4cs
February 2017 Highlights
• Disrup(on accelera(ng
• Demand to address mul(ple data types
• Demand for distributed processing and storage
• Cloud gaining trac(on
Data may include interac/on and observa/onal data fromInternet of Things (IoT) sensors
MemSQL 8
MemSQL
#1 Opera)onal Data Warehouse
Cri$cal Capabili$es for Data Warehouse and Data Management Solu$ons for Analy$cs
MemSQL 11
All Types of “Things”
Mul$ple data types
Massive scale with distributed systems
Always-on
Machine learning and predic$ve analy$cs
MemSQL 14
Ingest
• From batch to con-nuous
• High throughput
• Massively parallel
• Exactly-once seman-cs
MemSQL 17
Transform
• In-line, in real-,me
• Alter, enrich, score machine learning
• Use exis,ng models with PMML
• Add new models
MemSQL 18
Persist
• Full durability and availability
• Fast via memory op4miza4on
• Both real-4me and historical data
• Scale with distributed systems
• Robust security
MemSQL 19
Analyze
• ANSI SQL
• Sophis.cated query execu.on
• Code compila.on for fast queries
• Easy business intelligence integra.ons
MemSQL 20