Metabase Ljubljana
meetup #1
Metabase ❤

github.com/metabase/metabase
• Open source analytics tool

• 11k companies use us daily

• Largest user: Go-Jek with 4000 DAU

• Other notable users: N26, Under Armour, Swisscom,
Bang & Olufsen, NuBank, Convoy …

• Focus on UX and friendliness
Collect as much different
types of data as possible
Invest in a proper ETL
• Repeatability

• Visibility

• Extensibility

• Scalability

• Recoverability (don’t loose data, ever!)
ETL in a box
• segment.com

• Snowplow
Data warehouse topology
• Big fat denormalised tables, or 

• Star schema

• Use materialised views to tailor
the representation to your tools
Which DB?
• TL;DR: use Postgres

• Optimize for ease of ad-hoc querying

• Should be decently performant (waiting kills productivity) but is
unlikely to be the bottleneck

• Simple to deploy, connect to, and use

• Data validation/schemas where possible

• Sane handling of timezones/dates, numbers
Choosing an analytics tool
• Who will be using it?

• What are the deliverables?

• Size of data & scaling

• Will it be part of an existing proces or the core of a new one?
Deploying Metabase
• Switch app DB to Postgres (or MySQL)

• Docker

• Setup logging

• Upgrade to latest (but backup first)

• Ideally have 1 instance for all (if you need fine grained
access control talk to us about enterprise version)

• Have fun and fall in love! ❤
Coming soon
• Instance serialisation

• Reduced memory footprint

• More extensive support for joins (and a new query builder)
Questions
https://metabase.com

github.com/metabase/metabase

@sbelak

Metabase lj meetup

  • 1.
  • 2.
    Metabase ❤
 github.com/metabase/metabase • Opensource analytics tool • 11k companies use us daily • Largest user: Go-Jek with 4000 DAU • Other notable users: N26, Under Armour, Swisscom, Bang & Olufsen, NuBank, Convoy … • Focus on UX and friendliness
  • 3.
    Collect as muchdifferent types of data as possible
  • 4.
    Invest in aproper ETL • Repeatability • Visibility • Extensibility • Scalability • Recoverability (don’t loose data, ever!)
  • 5.
    ETL in abox • segment.com • Snowplow
  • 6.
    Data warehouse topology •Big fat denormalised tables, or • Star schema • Use materialised views to tailor the representation to your tools
  • 7.
    Which DB? • TL;DR:use Postgres • Optimize for ease of ad-hoc querying • Should be decently performant (waiting kills productivity) but is unlikely to be the bottleneck • Simple to deploy, connect to, and use • Data validation/schemas where possible • Sane handling of timezones/dates, numbers
  • 8.
    Choosing an analyticstool • Who will be using it? • What are the deliverables? • Size of data & scaling • Will it be part of an existing proces or the core of a new one?
  • 9.
    Deploying Metabase • Switchapp DB to Postgres (or MySQL) • Docker • Setup logging • Upgrade to latest (but backup first) • Ideally have 1 instance for all (if you need fine grained access control talk to us about enterprise version) • Have fun and fall in love! ❤
  • 10.
    Coming soon • Instanceserialisation • Reduced memory footprint • More extensive support for joins (and a new query builder)
  • 11.