What is a graph

database?
NOSQL CONCEPTS

Miriam Antona
Software engineer
Graph databases - overview
Treat data and its relationships with the same importance
Based on graph theory
branch of mathematics

studies graphs for modeling the relationships between objects

NOSQL CONCEPTS
Graph databases - parts of a graph
Vertices/nodes

Edges

NOSQL CONCEPTS
Graph databases - parts of a graph
Vertices/nodes
Edges/links/arcs

NOSQL CONCEPTS
Graph databases - nodes
Represent entities (users, cities, airports, employees...)
Have properties

NOSQL CONCEPTS
Graph databases - nodes
Represent entities (users, cities, airports, employees...)
Have properties

NOSQL CONCEPTS
Graph databases - edges
Connect the nodes
Define the relationships between the nodes

NOSQL CONCEPTS
Graph databases - edges
Connect the nodes
Define the relationships between the nodes

NOSQL CONCEPTS
Graph databases - edges
Connect the nodes

Define the relationships between the nodes

NOSQL CONCEPTS
Graph databases - types of edges
Directed:
Specific direction

NOSQL CONCEPTS
Graph databases - types of edges
Undirected:
No direction

The relationship is mutual

NOSQL CONCEPTS
Graph databases - queries
Traversing the graph
Examples:
Get all the users that Ben follows

Get when Carol started following Shui

Get the shortest path from one city to another

Path: set of nodes and edges across a graph

Query languages: Cypher, Gremlin, etc.

NOSQL CONCEPTS
Popular graph databases

NOSQL CONCEPTS
Let's practice!
NOSQL CONCEPTS
 Advantages and
limitations of graph
databases
NOSQL CONCEPTS

Miriam Antona
Software engineer
Advantages - flexibility
Can change as applications and industries change
Don't need to define the final final structure in advance

Can add/delete nodes, properties, and edges

NOSQL CONCEPTS
Advantages - performance
Doesn't need to perform joins
Joining can be potentially time-consuming

Follow edges from node to node

Simpler and faster

NOSQL CONCEPTS
Advantages - easy representation of the data
Similar structure to human thinking
Graph modeling very intuitive

Easily visualized

Facilitates understanding

NOSQL CONCEPTS
Advantages - horizontal scalability
It is possible
More difficult than in other NoSQL databases
Graphs are connected

Need to be distributed across multiple machines

NOSQL CONCEPTS
Limitations
Entity properties with extremely large values
BLOBs (Binary Large Objects): multimedia objects

CLOBs (Character Large Objects): collections of character data

Graph databases won't perform well

Bad practice

Use another database to store that information

Significant change for developers

New data modeling mindset

Learn Cypher, Gremlin...

NOSQL CONCEPTS
Let's practice!
NOSQL CONCEPTS
 When to use graph
databases
NOSQL CONCEPTS

Miriam Antona
Software engineer
Suitable cases - General information
Highly connected data

NOSQL CONCEPTS
Suitable cases - Social graphs
Social networks
Users and their relationships

Posts

Likes

...

Employees of a company
Projects

Relationships between the employees

NOSQL CONCEPTS
Suitable cases - Infectious diseases
Model:
Interactions between people

Contact events

Exposures

Help to understand transmission chains

NOSQL CONCEPTS
Suitable cases - Location services
Locations and distances
Optimize the routes for navigation
applications

Recommendations for nearby points of

interest:
restaurants

cinemas

hospitals

etc.

NOSQL CONCEPTS
Suitable cases - Fraud detection
Stop fraud in real-time:
Credit card fraud

E-commerce fraud

Money laundering

Model: individuals, credit cards, phone

numbers, devices, IP addresses...

Uncover suspicious patterns:

multiple users coming from the same IP

...

NOSQL CONCEPTS
Suitable cases - Real-time recommendations
Store buying history and browsing behavior
to recommend an item

Offer an item to a friend

Show unwatched movies based on similar

users

NOSQL CONCEPTS
Suitable cases - Networks
Storage of relationships between networks
and infrastructure elements
physical machines

virtual machines

applications

routers

switches

...

Alert in real-time about potential design

flaws

NOSQL CONCEPTS
Unsuitable cases
Disconnected data
Relationships between the data are not important

Applications that only perform general searches without a specific starting point
Are not optimized for those queries

Properties that contain extremely large values (BLOBs, CLOBs...)

NOSQL CONCEPTS
Let's practice!
NOSQL CONCEPTS
 Neo4j case study
NOSQL CONCEPTS

Miriam Antona
Software engineer
Neo4j - overview
Leader in graph database technology

420+ employees

Reveals and predicts how people,

processes, and systems are interrelated

NOSQL CONCEPTS
Neo4j graph platform
Suite of applications and tools
Interact and use graph data

NOSQL CONCEPTS
Neo4j graph platform - Graph database

Store and connect the data

NOSQL CONCEPTS
Neo4j graph platform - Graph data science library
Used for executing graph algorithms:
Path finding

Centrality

Community detection

Similarity

Link prediction

Node embeddings

Node classification

NOSQL CONCEPTS
Neo4j graph platform - Bloom

View and analyze data without any programming knowledge

NOSQL CONCEPTS
Neo4j graph platform - Cypher
Powerful query language
Inspired by SQL

Save and get data from the graph database

Multiple queries within the same transaction

CREATE (:User {name: 'Ann'})

NOSQL CONCEPTS
Neo4j graph platform - Connectors and Integrations
Facilitate the use of Neo4j in existing architectures
Neo4j Connector for Apache Spark

Neo4j Connector for Apache Kafka

Neo4j Connector for BI

Neo4j Labs Integrations

NOSQL CONCEPTS
Neo4j graph platform - Developer tools
Neo4j Desktop
Local Neo4j databases

Neo4j Browser
Interact with the graphs using a browser

Neo4j Sandbox
Learn about Neo4j, test personal ideas...

NOSQL CONCEPTS
Neo4j graph platform - Aura
Cloud Database Service
Run in the cloud without managing the
infrastructure

NOSQL CONCEPTS
Neo4j - drivers
Official drivers: C#, Java, Python, JavaScript, Spring...
Community drivers: R, PHP, Ruby...

NOSQL CONCEPTS
Neo4j - popular uses
Analytics and artificial intelligence
Fraud detection

Real-time recommendations

Knowledge graphs

Life sciences

Telecommunications

...

NOSQL CONCEPTS
Neo4j - customers

NOSQL CONCEPTS
Gousto - overview
Recipe box company
Delivers over a million meals each month

400+ staff

NOSQL CONCEPTS
Gousto - problem and solution
Problem: More choice implied difficulty to navigate
Solution: Use Neo4j
Internal recommendation system with recipes
subscriber's previous interactions with the menu

information on upcoming recipes

NOSQL CONCEPTS
Gousto - results
30% increase in the number of customers that select recommended recipes

Better cost control

NOSQL CONCEPTS
 Gousto - results
30% increase in the number of customers that select recommended recipes

Better cost control

1 https://neo4j.com/case-studies/

NOSQL CONCEPTS
Let's practice!
NOSQL CONCEPTS
 Congratulations!
NOSQL CONCEPTS

Miriam Antona
Software engineer
Chapter 1
Differences between NoSQL and relational databases
Main concepts of key-value databases

Advantages
simplicity

horizontal scalability

flexibility

Limitations
no complex queries

search just by key

NOSQL CONCEPTS
Chapter 1
Suitable cases
user-session information

user profiles and user preferences

shopping carts

...

Unsuitable cases
search a key based on its value

Case study: Editoo and Redis

reduction in downtime

higher performance

NOSQL CONCEPTS
Chapter 2
Basics of document databases
Advantages
flexibility

horizontal scalability

intuitive for developers

Limitations
care about the code

care about redundant data

NOSQL CONCEPTS
Chapter 2
Suitable cases
catalogs

event logging

user profiles

etc.

Unsuitable cases
very structured data

Case study: Shutterfly and MongoDB

improved its performance

NOSQL CONCEPTS
Chapter 3
Basics of column family databases
Advantages
flexibility

speed

horizontal scalability

Limitations
no multirow transactions

no joins

no subqueries

NOSQL CONCEPTS
Chapter 3
Suitable scenarios
event logging

Content Management Systems

time-series data

Unsuitable scenarios
need to change the queries very often

need complex queries and joins

don't deal with large amounts of data

Case study: Bigmate and Apache Cassandra

can handle concurrent millions of operations

NOSQL CONCEPTS
Chapter 4
Main concepts of graph databases
Advantages
flexibility

high performance

horizontal scalability

easiness of data representation

Limitations
extremely large objects (BLOBs, CLOBs...)

significant change for developers

NOSQL CONCEPTS
Chapter 4
Suitable situations
social graphs

infectious diseases

fraud detection

location services

etc.

Unsuitable situations
disconnected data

unimportant relationships between data

etc.

NOSQL CONCEPTS
Chapter 4
Case study: Gousto and Neo4j
better cost control

increased number of customers that select recommended recipes

NOSQL CONCEPTS
Thank you!
NOSQL CONCEPTS