Peak of Data & AI Encore: Scalable Design & Infrastructure

Unlocking FME Flow’s
Potential:
Architecture Design for
Modern Enterprises
The Peak of Data
and AI 2025

2025
The
Peak
of
Data
and
AI
Jean-Nicholas Lanoue
GIS Strategic Advisor
Since its inception in 1989, our GIS firm has been driven by a singular mission:
to help organizations unlock the full potential of their geospatial data
Théo Drogo
GIS Strategic Advisor

2025
The
Peak
of
Data
and
AI
Unlock the full potential of
FME Flow by making the
right architectural decisions,
whether you're deploying it
for the first time or scaling it
across the enterprise.

2025
The
Peak
of
Data
and
AI
In Scope (We will cover…)
▪Architecture Fundamentals
▪Express, Distributed or Fault-Tolerant?
▪Focus on Engines
▪Key architecture decision points
▪Real-world architecture examples
Out of Scope (20 minutes only 😉)
▪ FME Flow Hosted
▪ Infrastructure As Code
▪ FME Flow Administration concepts
▪ Containerization
▪ Licensing models
▪ And everything you won’t see in the next slides 👀

2025
The
Peak
of
Data
and
AI
Architecture Fundamentals
▪ 5 essential components
▪ Each play a specific role in execution & orchestration
▪ Applies to all deployments
… Only 5, but they scale in many ways
Web Application
Server
FME Flow Core FME Engine FME Flow
Database
FME Flow
System Share

2025
The
Peak
of
Data
and
AI
▪Express
3 main deployment options:
Express, Distributed, Fault-Tolerant

2025
The
Peak
of
Data
and
AI
▪Distributed

2025
The
Peak
of
Data
and
AI
▪Fault-Tolerant

2025
The
Peak
of
Data
and
AI
Engine deployment options
▪Same server, same network
▪Different server, same network
▪Different server, different network
Engine type in 1 word!
▪Standard engine : classic
▪Remote Engine: nomad
▪CPU-based engine: scalable
Focus on Engines

2025
The
Peak
of
Data
and
AI
Cloud
On-prem
Database
Linux
Windows
Hardware
ArcGIS
Web Server
Compatibility Performance
Other important considerations

2025
The
Peak
of
Data
and
AI
Key Architecture decision points
BUSINESS TECHNOLOGY
Time to
value
Criticality
Scalability
Compliance & Governance
Budget
Business
cases
Internal
expertise
Data
location
Security
Infrastructure
Availability
Technology
stack
Interoperability

2025
The
Peak
of
Data
and
AI
Real-life scenario - #1
Industry context
Local gov (population = 150k). GIS is a mature department.
B U S I N E S S T E C H N O L O G Y
Business cases
5–10 FME users
Low volume of geo and tabular data
Not mission-critical
Infrastructure
Fully on-premise environment
All apps deployed on internal VMs
Budget
Restricted budget
Cost-efficiency is a priority
Technology stack
Windows-only backend
ESRI 100% (Portal, Online, SDE, etc.)
SQL Server
Scalability
No short-term scaling needs
FME is not the corporate ETL
Data location
Internal file servers and database
Public APIs
Compliance &
Governance
Standard local gov compliance rules Availability No high availability or failover required
Internal expertise
5 GIS technicians, supported by IT
New to FME but tech-comfortable
Security
Deployed behind corporate firewall
SSL enforced for all services

2025
The
Peak
of
Data
and
AI
Distributed Architecture
o Database on SQL Server
o CPU Engine (1) on core server
o Standard Engine (1) on existing ArcGIS Server – for ArcGIS data integration
Industry context
Business cases
5–10 FME users
Infrastructure
Budget
Restricted budget
Restricted access to ESRI
licences
Technology stack
ESRI (Portal, Online, SDE)
SQL Server
Scalability
A few scaling needs for non-
spatial use
Data location
Public APIs
Compliance &
Governance
Standard local gov compliance
rules
Availability
No high availability or failover
required
Internal expertise
Security
Deployed behind corporate firewall

2025
The
Peak
of
Data
and
AI
Industry context
Business cases
5–10 FME users
Infrastructure
Budget
Restricted budget
Restricted access to ESRI licences
Technology stack
ESRI (Portal, Online, SDE)
SQL Server
Scalability
No short-term scaling needs
Data location
Public APIs
Compliance &
Governance
Standard logal gov compliance
rules
Availability No high availability or failover required
Internal expertise
Security
Deployed behind corporate
firewall
IT add-ons
o On prem VMs on Windows for the core, web server and CPU engine
▪ 6 vCPU + 32 Go RAM + 2 drives (OS vs App)
o Trusted SSL and SSO enabled
o Dedicated service account

2025
The
Peak
of
Data
and
AI
Real-life scenario #2
Industry context
Large gas company. FME is widely used across the organization..
Business cases
Large volume of data
Real time field collection
Mission critical
Infrastructure
ESRI Enterprise On-prem
Asset management on AWS
Ongoing cloud transition
Budget
FME subscription
Centralized FME budget (IT & GIS)
Technology stack
Hybrid OS
ArcGIS 100%
PostgreSQL
Scalability
Continuous growth on asset
Plan to expand to other geolocation
Data location Mix (cloud, on-site, SaaS)
Compliance &
Governance
Strong corporate governance rules
Industry compliance regulation
Availability
99.99%
( < 1 hour down time per year)
Internal expertise
Strong GIS admin team
10 years of FME experience
IT & GIS work together
Security
Security team involved
ISO27001 obligation

2025
The
Peak
of
Data
and
AI
Industry context
Business cases
Mission critical
Infrastructure
Budget
FME subscription
Technology stack
Hybrid OS
ArcGIS 100%
PostgreSQL
Scalability
Compliance &
Governance
Strong corporate governance rules
Availability
99.99%
Internal expertise
Strong GIS admin team
Security
ISO27001 obligation
Fault-tolerant Architecture
o Database on PosgreSQL on AWS
o Remote engines (2) on dedicated ArcGIS server only for licensing
o Standard Engines (4) for non-GIS data manipulation

2025
The
Peak
of
Data
and
AI
Industry context
Business cases
Mission critical
Infrastructure
Budget
FME subscription
Technology stack
Hybrid OS
PostgreSQL
Scalability
Compliance &
Governance
Strong corporate governance
rules
Availability
99.99%
Internal expertise
Security
ISO27001 obligation
IT add-ons
o AWS on Linux for the core, web server and standard engines
o On-site VMs on Windows for the remote engines
o SSL offloading of certificates through the loadbalancer

2025
The
Peak
of
Data
and
AI
“Why use one engine when you can
deploy five… in three networks…
and regret it later.”
— ChatGPT 4o on a Tuesday PM

2025
The
Peak
of
Data
and
AI
Thank You!
Curious to learn more?
Be sure to attend our next presentation at 3:00 PM!
Navigating FME Migrations:
The Essentials Before, During, and After
Ian Gagnon-Renaud
GIS Consultant
Laurent Phaneuf
GIS Consultant

Developing
Schemas with
FME & Excel
The Peak of Data
and AI 2025

2025
The
Peak
of
Data
and
AI
Mark
McCart
ETL Developer/FME Flow Admin
SAIC

1. Introduction
2. Solution Inspiration
3. Tools Used
4. Wrap-Up
Game Plan

Solution
Inspiration
Section 2

Explaining database schemas and
tracking field changes can be
challenging when the audience
doesn’t have a GIS or database
background.

2025
The
Peak
of
Data
and
AI
Excel is Familiar
● Easy to track schema mapping when
migrating data from one schema to another
● During warehousing tasks, a single Excel File
can track field name/type changes for multiple
tables
● Easy to re-order attributes, add notes and
other important annotation during the
migration process that can be used for final
documentation

2025
The
Peak
of
Data
and
AI
Typical Setup

2025
The
Peak
of
Data
and
AI
Esri and FME Tools
● Esri: Generate Schema Report Tool
● REST Endpoints - JSON
● FME: SchemaScanner Transformer

2025
The
Peak
of
Data
and
AI
I
=CONCAT(IFS(H2="String","fme_varchar("&I2&")",H2="Integer","fme_int32",H2="Double","fme_decimal(38,8)",H2="Date",
"fme_datetime",H2="Long","fme_int32",H2="Long Integer","fme_int32",H2="BigInt","fme_int64",H2="Short
Integer","fme_int16",H2="SmallInt","fme_int16",H2="GUID","fme_buffer"))

2025
The
Peak
of
Data
and
AI
Attribute Aliases
● Import ‘Attribute Values’

2025
The
Peak
of
Data
and
AI
Writer Configuration
● FGDB doesn’t need exist
● Set Coordinate System
● Use ‘Dynamic’ FC/Table Definition

2025
The
Peak
of
Data
and
AI
Writer Configuration
● FGDB doesn’t need exist
● Set Coordinate System
● Use ‘Dynamic’ FC/Table Definition
● Set Feature Class Name to
‘fme_feature_type_name’

2025
The
Peak
of
Data
and
AI
Next Steps
● Editor Tracking
● GlobalIDs
● Domains/Subtypes
● Attribute Rules

2025
The
Peak
of
Data
and
AI
Wrap-up:
● Developing and/or migrating schemas require detailed
documentation
● Spreadsheets are a useful tool for this type of task
● Have fun with FME, try something new with the
platform to see if it can solve one of your data problems

2025
The
Peak
of
Data
and
AI
Thank You
Mark McCart

Pixel Puzzler:
Extracting value
from images
Avineon Tensing
The Peak of Data
and AI 2025

2025
The
Peak
of
Data
and
AI
David
Eagle
Director of Service Delivery
Avineon Tensing

In 1981, Bill Gates said that…
“640kb of RAM ought to be enough
for anyone”
He was wrong!
https://en.wikipedia.org/wiki/Bill_Gates

In 1999, Dale Lutz said that…
“FME will NEVER do raster”
He was REALLY wrong!

2025
The
Peak
of
Data
and
AI
Number
of
supported
data
types
in
FME
1993 2000 2007 2010 2015 2017 2020 2025…
10
100
300
500+
GIS
CAD
Database
XML
Raster
3D
BIM
Web
Point Cloud
Cloud
Big
Data
IOT
Gaming
BI
Indoor
Mapping
AR/VR
Generative
AI
Cloud
Native
Tabular
The passage of time…

2025
The
Peak
of
Data
and
AI
You might be expecting…
Translate Transform Resample Reproject Mosaic
Compress Clip Tile Drape Calculate Shade Appearance

2025
The
Peak
of
Data
and
AI
But, actually…

Chat logs
Email archives
Video footage
Raw survey data
IoT Sensor data e.g. temp
Social media posts
Call centre logs
Sitesurvey photos
DARK DATA

Untapped Potential
Outdated or inaccurate
Costly – storage & environmental
Unnecessary backup durations
Legal and compliance issues
Hard to access
Awkward to search
Frustrating!
DARK DATA

DARK DATA
***PHONE STORAGE IS FULL***
Me: How can it be full already?
My camera roll:

2025
The
Peak
of
Data
and
AI
Google Pixel Phone - “Best Take”
iPhone - Classify people & pets!
https://blog.google/products/photos/how-google-photos-best-take-works/
https://support.apple.com/en-gb/guide/iphone/iph9c7ee918c/ios
Computer Vision

2025
The
Peak
of
Data
and
AI
Computer Vision
● If AI enables computers to think, computer vision
enables them to see!
● OCR – An early computer vision, developed in 1974
● That early tech now gives us real-time translation

2025
The
Peak
of
Data
and
AI
Computer Vision
● You have clever optics in your head
● But, computers have needed to be trained
● …and, fun fact… you’ve likely helped!

2025
The
Peak
of
Data
and
AI
CAPTCHA
“completely automated public Turing test to tell computers and humans apart.”

2025
The
Peak
of
Data
and
AI
Demo – Vision Comparison

2025
The
Peak
of
Data
and
AI
Prompt engineering
https://www.kaggle.com/whitepaper-prompt-engineering

Workspace Group – Case Study
https://l.ead.me/fmecasestudy

FME Hub – Verified Publisher

Computer Vision - R&D
● Utilising pre-trained models - Machine Learning
https://universe.roboflow.com/

2025
The
Peak
of
Data
and
AI
Demo – Orchestration

2025
The
Peak
of
Data
and
AI
UKPN – The Challenge
Identify the Cross-Section line and associated number from a map.
Review the ground layout for digitisation.

2025
The
Peak
of
Data
and
AI
UKPN – e2e Solution

2025
The
Peak
of
Data
and
AI
UKPN – e2e Solution (output)
Roboflow, Train the model: 1 day
Model Generation (Cloud Computing): 6-8 hrs
Object, Text Detection Success Rate: 75%-80%
Cost:
Roboflow $250 per month
Gemini $4

2025
The
Peak
of
Data
and
AI
Simon Green
Principal Technical Specialist
Siobhan Ryan
Junior Technical Specialist
Oliver Morris
Business Director
https://fme.safe.com/webinars/

2025
The
Peak
of
Data
and
AI
https://www.bbc.co.uk/programmes/m001wjf8

2025
The
Peak
of
Data
and
AI
“…and finally, a bit of fun!”

2025
The
Peak
of
Data
and
AI
Grab your phones!
Do:
● Launch the App
● Tap ‘browse file system’
● Snap something that belongs to you…
● Tap ‘Run’ and tell us how you were
categorised!!
Don’t:
● Take a photo of yourself or someone
else
● Upload a photo showing personal data
https://l.ead.me/FMEmyself

2025
The
Peak
of
Data
and
AI
Thank You
David Eagle
Avineon Tensing
deagle@avineon-tensing.com

Peak of Data & AI Encore: Scalable Design & Infrastructure

More Related Content

Similar to Peak of Data & AI Encore: Scalable Design & Infrastructure

More from Safe Software

Recently uploaded

Peak of Data & AI Encore: Scalable Design & Infrastructure