INNORIX in Practice, Fleet-Scale Collection

It started with collecting datafrom just 100 devices.

As a business grows, sites multiply and data diversifies.What will its data collection look like a few years from now?

Scenario
A fictional enterprise scenario in which data collection that began at a single site expands to countless devices across several countries.

2026ENTERPRISE GROWTH

Sites
86
Devices in scope
12,400
Daily data collected
28.4 TB
Countries
5
The fleet has grown 124x.How has data collection changed along the way?

86 sites and 12,400 devices appear, followed by one last question.

INNORIX in Practice

2022, The beginning2022.03

THE BEGINNING, March 2022

The first 100 connected devices

The company started collecting data so it could review, in one place, the records of the devices running at a single site.

Back then, the devices produced only a few kinds of files.

All it took was gathering operation logs and work results on a company server after the day’s work was done.

The team connected the devices, specified where the files lived, and set up a collection job to run every day at 11 PM.

At the start, there was no reason to build a complex data collection environment.

MARCH 2022

One site, one collection server

Sites
1
Devices
100
Daily collection
80 GB

SOURCE DEVICES

Site-001, 100 devices

COLLECTION DESTINATION

Enterprise server

  • /field-data/
  • 2022-03-01/
  • Device-001/
  • Device-002/
  • ...
  • Device-100/

INNORIX STUDIO, FLOWS

Daily field collection

Source devices
Site-001, 100 devices
Source files
/logs/, /results/
Destination
Enterprise server
Schedule
Every day at 23:00
Save rule
Separate folders by device and date
Files created by 100 devicesare collected directly to the company server every day.

Live contextNow

2022.03

Sites
1
Devices in scope
100
Daily data collected
80 GB

COLLECTION METHOD

Direct collection

CURRENT TASK

First scheduled collection job

COMPARED WITH DAY ONE

1x

THE FIRST EXPANSION, February 2023

New sites opened their doors.

A year later, the company was running five new sites.

The fleet grew from 100 devices to 680.

The new sites produced the same kinds of operation records and work results as the original one.

Instead of building a new data collection program, the company connected the new sites’ devices and added them to the existing collection job.

Collection at the first site kept running exactly as before.

FEBRUARY 2023

Five new sites join the original one.

  • Site-001Existing

    100 devices

  • Site-002New

    120 devices

  • Site-003New

    80 devices

  • Site-004New

    140 devices

  • Site-005New

    110 devices

  • Site-006New

    130 devices

INNORIX STUDIO, DEVICES

Connected devices

Device groupDevicesConnectionCollection
Site-001100ConnectedDaily collection
Site-002120ConnectedDaily collection
Site-00380ConnectedDaily collection
Site-004140ConnectedDaily collection
Site-005110ConnectedDaily collection
Site-006130ConnectedDaily collection
In scope last year
100
In scope now
680
The number of sites grew,but the existing collection job didn’t have to be rebuilt.

Live contextNow

2023.02

Sites
6
Devices in scope
680
Daily data collected
620 GB

COLLECTION METHOD

Direct collection

CURRENT TASK

Connecting devices at new sites

COMPARED WITH DAY ONE

6.8x

August 2023

Devices in the same company don’t all produce the same files.

The company began bringing in devices that did different kinds of work.

Some devices produced a steady stream of small text files, while others generated images and measurement data running to hundreds of MB.

Some devices needed their results captured in near real time; for others, collecting records once a day was enough.

Applying the same collection rules to every device no longer made sense.

The company started grouping devices not only by site, but also by the role they actually play.

AUGUST 2023

Each device produces different data and needs a different collection time.

  • Operation devices360 devices

    runtime.logoperation.csv

    Collection time

    Daily at 23:00

  • Inspection devices120 devices

    inspection.csvoriginals/

    Collection time

    After work completes

  • Measurement devices140 devices

    measurements.binsensor.csv

    Collection time

    Every hour

  • Site management systems60 devices

    daily_summary.csvwork_history.json

    Collection time

    Daily at 22:00

INNORIX STUDIO, DATASETS & FLOWS

Collection by device role

DatasetDevicesCollection rule
Operation records360Daily
Inspection records120On completion
Measurement data140Hourly
Management records60Daily
Even devices at the same sitecan take part in different collection jobs.

Live contextNow

2023.08

Sites
6
Devices in scope
680
Daily data collected
620 GB

COLLECTION METHOD

Collection by device group

CURRENT TASK

Separating targets by device role

COMPARED WITH DAY ONE

6.8x

WHEN SCALE CHANGES, March 2024

Thousands of devices finish work at about the same time.

The company now ran 24 sites nationwide.

2,400 devices were generating and sending in about 3.8 TB of data a day.

The challenge wasn’t just the total volume collected each day.

With many sites across the country finishing work at around the same time, collection requests began piling up in specific hours.

A single collection server could still handle it, but the company had to plan for the devices and data still to come.

It set up a grid collection environment that spreads collection across several collection servers.

MARCH 2024

A day’s data converges on a few hours.

COLLECTION REQUESTS BY HOUR, EXAMPLE

Peak collection hours

Collection requests rise as sites nationwide wrap up the day

DIRECT COLLECTION

Single-server collection

Collector

Handled by one collection server

Keep direct collection when the environment and capacity are sufficient.

GRID COLLECTION

Grid collection

Collector 1Collector 2Collector 3Collector 4

Spread across several collection servers

Built to keep up as collection requests and data volumes grow.

GRID COLLECTION, EXAMPLE WORKLOAD

  • Collector 01612
  • Collector 02584
  • Collector 03603
  • Collector 04601

Example distribution of collection work across 2,400 devices

As collection targets grow,the environment that handles collection can grow with them.

Live contextNow

2024.03

Sites
24
Devices in scope
2,400
Daily data collected
3.8 TB

COLLECTION METHOD

Grid collection

CURRENT TASK

Distributing the collection load

COMPARED WITH DAY ONE

24x

September 2024

Not every record had to come back to headquarters.

As data flowed in from sites across the country, new requirements emerged.

Headquarters needed work results and key operating information from every site.

But detailed video data and raw device records were used mostly by staff in each region.

Collecting every file at headquarters didn’t match how the company actually used its data.

The company separated the data headquarters needed from the data used regionally, and changed its collection setup so some of it is kept in regional storage.

SEPTEMBER 2024

Data is collected where it’s actually needed.

HEADQUARTERS COLLECTION

24 sites nationwide

Work results, key operation records, summary data

HQ storage620 GBCollected daily

REGIONAL COLLECTION

  • Region A

    Regional storage

    8 sites, 780 devices

    1.04 TB

    Detailed operations, raw inspection data

  • Region B

    Regional storage

    9 sites, 940 devices

    1.28 TB

    Detailed operations, raw inspection data

  • Region C

    Regional storage

    7 sites, 680 devices

    860 GB

    Detailed operations, raw inspection data

DAILY DATA DESTINATIONS

3.8 TB

  • HQ storage620 GB
  • Region A1.04 TB
  • Region B1.28 TB
  • Region C860 GB
Gathering all data in one placeis not the same as keeping it where it belongs.

Live contextNow

2024.09

Sites
24
Devices in scope
2,400
Daily data collected
3.8 TB

COLLECTION METHOD

Direct and site-based collection

CURRENT TASK

Setting up regional storage

COMPARED WITH DAY ONE

24x

THE BUSINESS KEEPS GROWING

When the first collection job was set up,no one expected today’s scale.

20222023202420252026
Sites16244886
Devices1006802,4006,20012,400
Daily collection80 GB620 GB3.8 TB11.6 TB28.4 TB
Sites
86x
Devices
124x
Daily data collected
355x

No one can know in advance how fast a business will grow.But it does need an operating environment ready for each new scale.

INNORIX in Practice

2022, The beginning2022.03

BEYOND ONE COUNTRY, February 2025

Data now has to be collected overseas, too.

Moving beyond its home market, the company began running sites overseas.

It now operated 6,200 devices across 48 sites in 3 countries.

Each country started and ended its workday at a different time.

Network conditions and storage locations differed too, and company policy required some data to stay within its country.

Collection rules built for domestic sites couldn’t simply be copied to every overseas site.

The company set up collection to match each region’s local work schedule and data governance rules.

FEBRUARY 2025

The company’s day no longer runs on a single time zone.

  • Country A, UTC+9

    23:00

    Local collection start

    36 sites

    4,200 devices

    Actual run time

    14:00 UTC

  • Country B, UTC+7

    23:00

    Local collection start

    8 sites

    1,400 devices

    Actual run time

    16:00 UTC

  • Country C, UTC−5

    23:00

    Local collection start

    4 sites

    600 devices

    Actual run time

    04:00 UTC

COLLECTION SCHEDULE, UTC

Collection in Country C starts at 04:00 UTC the next day.

COLLECTION DESTINATION POLICY

Data typeDestination
Company-wide work resultsHQ storage
Regional operation recordsStorage in that region
Data that must stay in-countryDesignated storage in that country
Large regional raw dataRegional Object Storage
Even when collection is managed company-wide,not all data has to move to the same place at the same time.

Live contextNow

2025.02

Sites
48
Devices in scope
6,200
Daily data collected
11.6 TB

COLLECTION METHOD

Regional collection

CURRENT TASK

Overseas sites and local schedules

COMPARED WITH DAY ONE

62x

August 2025

Devices once run by another company joined the fleet.

As part of its expansion, the company acquired another company’s operations team and devices.

The acquired sites used different operating systems, software and file storage conventions.

Some devices ran on industrial Linux; others ran on Windows.

Replacing every device’s operating system or business software was not a realistic option.

The company kept the existing environments in place and connected supported devices to data collection first.

AUGUST 2025

One company now, but device environments still differ.

EXISTING OPERATIONS

Original company environments

  • WindowsOperation devices

    Existing collection kept as is

  • UbuntuMeasurement systems

    Existing collection kept as is

  • Rocky LinuxSite servers

    Existing collection kept as is

NEW OPERATIONS

Acquired company environments

  • Embedded LinuxIndustrial devices

    New connection and collection rules

  • DebianSite control systems

    New connection and collection rules

  • Windows ServerData management servers

    New connection and collection rules

INNORIX STUDIO, DEVICES

New device groups

Device groupEnvironmentCollection source
Acquired-OperationsEmbedded Linux/var/log/device/
Acquired-ControlDebian/opt/control/output/
Acquired-ServersWindows ServerD:/operation/results/
Collecting data doesn’t requiremaking every site’s environment the same.

Live contextNow

2025.08

Sites
48
Devices in scope
6,200
Daily data collected
11.6 TB

COLLECTION METHOD

Multiple collection environments

CURRENT TASK

Integrating new device environments

COMPARED WITH DAY ONE

62x

MORE THAN AN IT OPERATION, April 2026

Collection now serves many teams across the company.

The company now runs 86 sites and 12,400 devices in 5 countries.

When collection began, its main purpose was to give the IT operations team the device records it needed.

Over time, though, Quality Management, Technical Support, R&D and Business Operations started needing field data too.

Even for data from the same devices, each team needed different files, collection times and storage locations.

Data collection was no longer a job for just one department.

APRIL 2026

One field, different data needs for every team

SHARED DEVICE ENVIRONMENT

86 sites, 5 countries

12,400 devices

  • IT Operations

    Data needed
    Device operation logs, error records
    When
    Hourly and whenever an error file is created
    Destination
    Operations storage
  • Quality Management

    Data needed
    Inspection results, raw inspection data
    When
    After work completes
    Destination
    Quality storage
  • Technical Support

    Data needed
    Detailed records of selected devices
    When
    On investigation request
    Destination
    Technical workspace
  • R&D

    Data needed
    Test results, measurement data
    When
    After each test
    Destination
    Research storage
  • Business Operations

    Data needed
    Work results by site
    When
    Daily, after work ends
    Destination
    Business storage
  • Data Analytics

    Data needed
    Datasets for analysis
    When
    Before each scheduled analysis
    Destination
    Analytics storage

INNORIX STUDIO, FLOWS

Collection workflows

FlowOwnerExecution
Operational logsIT OperationsScheduled
Quality inspectionQuality ManagementEvent
Technical investigationTechnical SupportOn demand
Research datasetsR&DEvent
Daily business recordsBusiness OperationsScheduled
Analytics datasetsData AnalyticsScheduled
Each new use for the datadoesn’t mean rebuilding the device environment.

Live contextNow

2026.04

Sites
86
Devices in scope
12,400
Daily data collected
28.4 TB

COLLECTION METHOD

Purpose-based collection

CURRENT TASK

Collection set up per team

COMPARED WITH DAY ONE

124x

OPERATING AT A DIFFERENT SCALE, September 2026

Hundreds of collection exceptions occurred again today.

With 100 devices, someone could check any device that had a problem by hand.

But in an environment running 12,400 devices, all kinds of situations come up every day.

Some devices dropped off the network, others were under scheduled maintenance. Some had not created their files yet, and one storage location was running out of space.

Not every device was running normally all the time.

What mattered was not hoping these situations would never happen, but continuing whatever collection was possible when they did, and clearly identifying what remained unfinished.

SEPTEMBER 2026, 23:00

Not all 12,400 devices are in the same state.

DAILY COLLECTION, INITIAL RESULT

Initially incomplete560

12,400 devices

  • StatusDevices
  • Collected normally11,840
  • Network disconnected248
  • Maintenance and reboot164
  • Waiting for target files96
  • Affected by low storage52

COLLECTION RECOVERY

More devices become collectable over time.

Collected normally
11,840
Collected after recovery
534
Incomplete
26

INNORIX STUDIO, RUNS

Remaining collection targets

DeviceSiteLast conditionStatus
Device-08421Site-042DisconnectedPending
Device-10284Site-068MaintenancePending
Device-11036Site-074File not readyPending
Device-11862Site-081Storage unavailablePending

Some of the 26 incomplete devices

Instead of rechecking all 12,400 devices,focus on the devices and files not yet collected.

Live contextNow

2026.09

Sites
86
Devices in scope
12,400
Daily data collected
28.4 TB

COLLECTION METHOD

Exception recovery and recollection

CURRENT TASK

Checking incomplete devices and files

COMPARED WITH DAY ONE

124x

November 2026

Checking collection across more than 10,000 devices.

At 8:30 AM, the company’s data operations manager reviews the collection that has run since the previous day.

Unlike the days of 100 devices, collection jobs for countless sites and many teams now run at the same time.

The manager doesn’t check every device’s files one by one.

Instead, they review results by country, by site and by collection job, and check for any files or devices left incomplete.

When needed, they open a device’s detailed collection history to see which files were collected and which remain.

NOVEMBER 2026, 08:30

From the company-wide view down to the records of a single device.

INNORIX Studio

Collection runs

Devices in scope
12,400
Files collected
8.6M
Data collected
28.4 TB
RegionDevicesCompletedPendingData
Country A5,2005,191910.8 TB
Country B2,8002,79376.2 TB
Country C1,8001,79644.6 TB
Country D1,6001,59643.8 TB
Country E1,00099823.0 TB

COLLECTION HISTORY, INITIAL DEVICES

Site-001

The first device group collected2022.03

Current devices100

FlowLast runStatus
Daily field collection2026.11.15 23:00Completed
The very first collection job, started four years ago,still runs today within the company-wide collection environment.

Live contextNow

2026.11

Sites
86
Devices in scope
12,400
Daily data collected
28.4 TB

COLLECTION METHOD

Direct, grid and site-based

CURRENT TASK

Reviewing company-wide results

COMPARED WITH DAY ONE

124x

2022 — 2026

It started as gathering files.Now it’s a data operation that serves the whole business.

2022, THE BEGINNING

100 devices

Sites
1
Countries
1
Daily collection
80 GB
Collection setup
Single server
Purpose
Keeping operation records
Teams served
IT Operations

2026, ENTERPRISE OPERATIONS

12,400 devices

Sites
86
Countries
5
Daily collection
28.4 TB
Collection setup
Direct, grid and site-based
Purpose
Company-wide data use
Teams served
6 key teams

As the company grew, it expanded its collection environment step by step.There was no need to build everything on day one.

THE NEXT EXPANSION

2027 — More new sites are on the way.

The company plans to add 12 sites next year.

The new sites will also bring in types of devices it has never run before.

About 2,400 devices will join the collection environment, and how much data the new sites produce will only be known once they are up and running.

No company can predict everything that lies ahead.

But it is ready to connect new devices, configure collection targets and storage locations, and expand collection servers and regional environments as needed.

Live contextNow

2027, planned

Sites
98
Devices in scope
14,800
Daily data collected
TBD

COLLECTION METHOD

Expansion planned

CURRENT TASK

Preparing new sites

COMPARED WITH DAY ONE

148x

2027, PLANNED EXPANSION

The company is still growing.

  • First site
  • Operating sites
  • Planned sites
Planned new sites
12
Planned new devices
2,400
Expected total devices
14,800

NEW SITE ONBOARDING

Preparation taskApproach
Connect devices at new sitesSet up new device groups
Check file types per deviceApply existing or new Datasets
Decide when to collectMatch local work schedules
Decide where data is storedAssign HQ or regional storage
Review collection capacityExpand the collection environment as needed
Adding new sitesis now routine work in the company’s data operation.

BUILT FOR A BUSINESS THAT KEEPS GROWING

No one can know from the starthow far a business will grow.

  1. 2022100 devices
  2. 2023680 devices
  3. 20242,400 devices
  4. 20256,200 devices
  5. 202612,400 devices

A job that began with the first 100 connected devicesnow runs across 12,400 devices.

Sites multiply, devices diversify, and the uses of data change.A business must be able to build the collection environment each stage of growth requires.

Today’s data collectionshould never limit tomorrow’s growth.

THE TECHNOLOGY BEHIND THIS STORY

Fleet-Scale Collection

From the devices at a single site to countless industrial devices spread across countries, INNORIX collects the data your business needs.

Start with direct collection, then add grid and site-based collection as your scale and operating environment grow.

Explore the product