An Architectural Overview of 3 IoT Use Cases

1
An Architectural Overview of 3 IoT Use Cases

• 14th June 2017

2
Agenda

• Review of what an end-to-end IoT platform needs
• IoT Data Characteristics
• Lessons Learned
• Use Cases
• Use case 1 BBOXX
• Use case 2 Spiio
• Use case 3 tado
• QA

3
Who we are

• Mark Herring, CMO

• Michael DeSa, SW Engineer

4
InfluxData Overview
Founded in 2013
Delivering a modern open source platform for
metrics and events
Guiding principles Developer Happiness Ease of
Development Scale Out Time to Value
Results 70,000 Active Servers 10,792 GitHub
stars 300 Customers (Cloud and Enterprise
Offering)
5
InfluxData Customers
6
A Review of an End-to-End IoT platform
7
IoT Data Characteristics

• IoT Data is Time Data
• IoT Data is Streaming Data
• IoT Data is Real-Time

8
A Review of an End-to-End IoT platform
9
Lessons Learned

• Lesson 1 Dont start with the wrong data
  platform architecture
• Lesson 2 Dont settle for first generation Time
  Series Databases
• Lesson 3 Use an IoT Data PLATFORM not just a
  database

10
Lesson 1 Dont start with the wrong data
platform architecture

• The typical path we see with IoT projects
• Start with MySQL
• Try HBase or Cassandra
• Try Elasticsearch
• Try MongoDB

11
Lesson 1 Dont start with the wrong data
platform architecture

• Conclusion Adopt a Time Series Database Database

12
Lesson 2 Dont settle for first generation Time
Series Databases
InfluxDB outperformed OpenTSDB by 5.0x when evaluating data ingestion performance 16.5x better on-disk compression 4.0x faster query performance

13
Lesson 2 Conclusion
Use a Modern Time Series Database
The Modern Engine for Metrics and Events
14
Lesson 3 Use an IoT Data PLATFORM not just a
database
visualize
DASHBOARDS
notify
Collect, normalize, correlate, and aggregate
metrics and events from over 100 data sources
Analyze, store and manage time series data. Use
machine learning libraries for anomaly detection.
Visualize interesting trends, detect events using
time series functions, automate your entire
system.
CUSTOMERSDECISION-MAKERS
automate
MACHINE LEARNING
ARCHIVE
15
Conclusion Use InfluxData as the IoT Data
Platform
visualize
DASHBOARDS
Open Source Core
notify
Collect, normalize, correlate, and aggregate
metrics and events from over 100 data sources
Analyze, store and manage time series data. Use
machine learning libraries for anomaly detection.
Visualize interesting trends, detect events using
time series functions, automate your entire
system.
CUSTOMERSDECISION-MAKERS
Enterprise Features
automate
MACHINE LEARNING
ARCHIVE
16
Use Case 1 BBOXX
17
David McLeanLead Developer at BBOXX

• IOT USE CASE BBOXX

We analyze over 70,000 hours of data every night,
half a billion data points, to produce alerts for
our technicians. Having this real-time data in
the cloud makes it possible to identify trends,
usage patterns even detect problems before they
exist!

• BBOXX develops solutions to provide affordable,
  clean energy to off-grid communities in the
  developing world.
• They continuously monitor 85,000 solar based
  systems providing insights into their
  customer-usage patterns and anomaly detection.

• What are they trying to achieve with Time Series
  data
• Scale support their goals to go from 85,000
  units to 20 million units by 2020
• Growth achieved thru expand pricing plans based
  on data captured
• Customer Satisfaction query real-time data fast

18
BBOXX Data Architecture
19
Architecture BBOXX

• Datastore
• InfluxDB
• Analytics/Visualization
• Built in-house, backed by InfluxDB queries
• Aggregation/Correlation
• Stored back into InfluxDB

20
Lessons BBOXX

• Lesson 2 Use a modern Time Series Database
• Using a modern Time Series Database allows BBOXX
  to scale from 85,000 units to 20 million units by
  2020 without re-architecting their data pipeline.

21
Use Case 2 Spiio
22
Jens-Ole GraulundCTO at Spiio

• IOT USE CASE Spiio

As more people populate cities and miss nature,
nature is moving to the city. But for nature
cities to be a reality, we need to understand
greenery performance from data. Thats why Spiio
is using InfluxData - it is the tech enabler for
our vision bridging the gap between things and
people.

• Spiio uses its sensor-based platform to give
  clients a full view of green wall installations
  anytime, anywhere. Using real-time analytics,
  clients can understand their plants condition,
  share insights across their org, make data-driven
  decisions to boost maintenance efficiency
  improve green wall design.

• What are they trying to achieve with Time Series
  data
• Customer Efficiency - provide insights to their
  customers on current future plant conditions,
  allowing them to focus on building out more plant
  installations
• Avoid guesswork - track the impact of factors
  that influence plant performance

23
Spiio Data Pipeline
24
Architecture Spiio

• Datastore
• InfluxDB
• Analytics/Visualization
• Chronograf
• Custom in-house app
• Aggregation/Correlation
• Kapacitor
• Custom in-house tool

25
Lessons Spiio

• Lesson 3 Use an IoT Data Platform not just a
  database
• Utilizing an IoT Data Platform allowed Spiio to
  focus on building out their application, rather
  than spending engineering effort on the logistics
  of managing, building, and maintaining a custom
  time series data platform.

26
Use Case 3 tado
27
Michal KnizekHead of Server Development at tado

• IOT USE CASE

While testing other solutions, we tried our
current production loads against InfluxDB found
that it exceeded our needs. We now use it to
generate our new user report - InfluxDB is so
cool very fast even with this increase in use,
we expect higher data loads that we know InfluxDB
can handle.

• tado has been connecting heating systems with
  the internet and making their control even
  smarter. In June 2015, tado augmented its
  service with a Smart AC Control that enables
  consumers to intelligently control air
  conditioners.

• What are they trying to achieve with Time Series
  data
• Scale against their current load of over 1
  million devices
• Customer Satisfaction - eliminate user report
  response time issues - used to engage customers
  by providing instant access to daily insights
  full years view of their home temperatures

28
tado Data Pipeline
29
Architecture tado

• Datastore
• Originally MySQL
• Previously Custom in-house
• Currently InfluxDB
• Analytics/Visualization
• User facing app

30
Lessons tado

• Lesson 1 Start with the right data platform
  architecture
• Initially, tado started with MySQL as their time
  series database. This worked when they had 10,000
  customers, but they quickly began to hit issues
  scaling beyond that.
• Since switching to InfluxDB, theyve been able to
  handle millions of customers.

31
Conclusion
32
Using the Right Tool for the Right Project

• Lesson 1 Start with the right data platform
  architecture
• Lesson 2 Use a modern Time Series Database
• Lesson 3 Use an IoT Data Platform not just a
  database

33
Next Steps

• Visit InfluxData.com
• Download the paper on Choosing the right IOT Data
  Platform
• Ask Questions on the community
  community.influxdata.com