[ Index ]

QL Resources Migration (AWS MAP)

Led an AWS Migration Acceleration Program engagement moving QL Resources’ agricultural and HR reporting, more than 50 servers and their databases, onto AWS.

Client
QL Resources
Role
Project lead
Period
2025
Stack
Amazon EC2 · Amazon Aurora · AWS Database Migration Service

The problem

QL Resources, a Malaysian agro-food group spanning marine products, livestock farming and palm oil, ran its agricultural and HR reporting on more than 50 servers in its own data centre, across three database engines. Growing meant buying hardware: every bit of extra capacity had to be bought and installed, sized for the busiest day and idle the rest of the time. Keeping it all running was a job in itself, with patching, hardware refreshes, backups and failed parts all landing on the IT team, and if that one data centre went down, the reporting went with it.

The solution

I led the move under the AWS Migration Acceleration Program:

  • A landing zone and a private network, in place before a single server moved. The landing zone settled accounts, access and a security baseline up front, and the network kept the estate working while cutover split it between the data centre and AWS.
  • More than 50 servers moved to Amazon EC2 in planned waves. Each one was right-sized to how it was actually used, not the specification it was bought with, so years of spare headroom did not follow it into the monthly bill.
  • The databases moved to Amazon Aurora with AWS Database Migration Service, including a translation of SQL Server’s schema and stored procedures.
  • Backup, disaster recovery and monitoring were part of the job, so the estate could be run the morning after it landed, not just delivered to AWS.

One engine was not like the others

The reporting ran on three database engines. PostgreSQL and MySQL each had a matching edition of Amazon Aurora to move to, so the real risk there was the cutover itself. SQL Server was a different story. Aurora has no SQL Server edition, so its schema and stored procedures had to be rewritten for a new engine. And a report that still runs afterwards is not proof that it returns the same figures it did before.