Webanite: modernising a 248-million-row WordPress platform on AWS

The challenge

Webanite's production stack ran on MySQL 5.7 and WordPress 4.9. AWS moved MySQL 5.7 into paid Extended Support, adding an extra $2,100/month (Year 3 pricing). The data was too big for a simple dump-and-restore:

• 3 RDS databases and 85 GB of WordPress files;
• a 248 million-row order_statement table and several 51 million-row tables;
• live API traffic that couldn't take a long outage.

Our approach

1. Separate AWS account with a complete new environment: PHP 8.2 on ECS Fargate, HTTPS via Cloudflare.
2. Upgrades: MySQL 5.7 → 8.0, WordPress 4.9 → current, with all breaking queries fixed (GROUP BY, charset) and every API endpoint validated.
3. Python sync engine (51 tests) with a live dashboard. A lightweight shadow changelog replaced the timestamp-index approach, so changes are captured in real time without production downtime.
4. Blue-green cutover plan: the new environment becomes production with a Cloudflare DNS flip, with instant rollback.
5. Along the way: fixed a live Redis outage caused by a VPC peering route conflict, built a production monitoring TUI and scheduled index builds in Sunday low-traffic windows.

The result

• New environment continuously in sync with production across all four high-volume tables with “effectively zero drift” (12 Jun 2026).
• Recovered a gap of ~30,000 rows missing from the original snapshot of the 248M-row table.
• Planned final switchover: under 30 minutes of downtime.
• Removes exposure to the $2,100/month MySQL 5.7 Extended Support charge once cut over.