Home
/
Case studies
/
Edge MySQL tables replicated to Azure with no maintenance
Integration and data

Edge MySQL tables replicated to Azure with no maintenance

Client
A multi-site manufacturing group
Industry
Multi-industry
15 min

change-check interval

A multi-site manufacturing group needed exact Azure copies of its local Ignition tables without touching edge performance. Plantformance built a change-data-capture pipeline on MySQL internals that checks marked tables every 15 minutes, compresses changes about 1:50 into Parquet files, and ships them through NiFi. Turning on sync for a table now takes a 2-minute change request.

Key results

  • 15 min change-check interval
  • 1:50 typical compression ratio
  • 2 min to build a sync change request

The problem

Ignition data usually lives locally and serves many activities. It is often stored in the local edge MySQL database, the default on FactoryBrain, with excellent availability, speed and consistency. The enterprise needs that same data for master-data sync, performance, aggregated data and analytics. The client asked for a solution with no maintenance that would replicate any local schema or table into its Azure cloud.

What it looked like before

Some data was already streamed, which gave a rough view of local data in the cloud. Streaming did not meet the reliability goal: a perfectly synchronized copy in the cloud of what exists locally.

How Plantformance solved it

Plantformance built a change-data-capture solution based on MySQL internals, with no local data duplication and no performance impact. Every 15 minutes it checks the tables marked for sync and writes the changes, inserts, updates and deletes, into heavily compressed Parquet files, usually 1:50. The files are buffered locally and centrally with NiFi, then shipped to Azure Blob Storage with a strict naming scheme so data scientists can rebuild tables on the fly. Checks detect missing or corrupted data at every stage and allow a resync.

What changed

Turning on sync for a table is a change request that takes 2 minutes to build. Implementation takes Plantformance 5 minutes.

Results

  • Local data stays at the edge for speed, reliability and performance for plant teams
  • Operations do not depend on the cloud
  • Enterprise teams get a synchronized copy of local tables in Azure
  • Changes compressed about 1:50 before shipping
  • Missing or corrupted data is detected and can be resynced

Plantformance products and services used

FactoryBrain, Professional Services

Frequently asked questions

Does the cloud sync slow down the edge database?

No. Plantformance built the change-data-capture solution on MySQL internals, with no local data duplication and no performance impact. Local data stays at the edge for speed, reliability and performance for plant teams, and operations do not depend on the cloud.

How is missing or corrupted data handled?

Changes are buffered locally and centrally with NiFi, then shipped to Azure Blob Storage with a strict naming scheme. Plantformance added checks that detect missing or corrupted data at every stage and allow a resync, so the cloud copy stays synchronized with local tables.

How long does it take to add a table to the sync?

Turning on sync for a table is a change request that takes 2 minutes to build, and implementation takes Plantformance 5 minutes. From then on, the pipeline checks the table every 15 minutes and writes inserts, updates and deletes into compressed Parquet files.

Related Ignition case studies

Planning a similar project? Talk to a Plantformance Ignition engineer or browse all 37 Ignition case studies.

Planning a similar Ignition project?

Tell us about your plants and your Ignition setup. A Plantformance engineer will show you how a project like this would work for you.