FRFaraz Rohaila

Case study · Prop trading firm

A 30 GB WordPress store buckling under its own weight. Fixed by re-architecting, not rebuilding.

How a prop firm running WooCommerce, a trading CRM, and manual crypto payment processors moved from a fragile single-server setup to a stable AWS architecture with automated reconciliation.

  1. Single server bottleneck
  2. Workload separation
  3. Reliable platform without rebuild
CLOUDFRONT + WAFAPPLICATION LAYERWEBADMINWORKERSRDS MULTI-AZ / REDIS / S3AUDIT + RECONCILIATIONNIGHTLY · IMMUTABLE · QUERYABLE
30 GB+migrated, zero data loss
90 → 1days to detect anomalies
Near-zerodowntime during migration

02 / The architecture shift

We didn't rebuild. We separated responsibilities.

The whole case in one move: everything that used to depend on one machine now runs in the environment it needs.

Before · one machineAfter · separated environments
One server doing everything Separated workloads
Files stored locally Cloud storage
Admin slows customers Independent admin environment
Manual reconciliation Nightly validation
Editable records Immutable records

03 / Project snapshot

A platform problem disguised as a WordPress problem.

The client believed they had outgrown WordPress. The actual issue was running every workload on one machine.

Client

Proprietary trading firm

Funded-account challenge platform

Existing stack

WordPress + WooCommerce

Trading CRMCrypto payment processors

Scale

30 GB+

Database + years of media

Business risk

Quarterly manual reconciliation

90-day detection delay

Constraint

No rebuild. No replatform.

No change to order IDs.

Role

Architecture review

Migration planningTechnical delivery leadership

04 / Architecture

Keep what works. Change what doesn't.

The solution was not rebuilding the application. WordPress, WooCommerce, plugins and business rules stayed. The infrastructure changed underneath them, one layer at a time, in the order a request travels.

  1. Request

    Customers, admins and integrations enter through one controlled edge.

  2. 01 / Edge

    Traffic and protection

    Reduce origin load and protect customer-facing traffic.

    Route 53CloudFrontAWS WAFALB
  3. 02 / Application

    Application separation

    Customers, admins and scheduled processes run in separate services so they stop competing.

    ECSECRContainersWorkers
  4. 03 / Data

    Data reliability

    Remove the database as a single point of failure and take media off the application servers.

    RDS MySQLMulti-AZRedisS3
  5. 04 / Audit

    Financial reconciliation

    Turn quarterly investigation into nightly validation on immutable records.

    S3 Object LockGlueAthenaQuickSight
  6. Operations

    Make problems visible before customers report them.

    CloudWatchCloudTrailSecrets Manager

05 / Migration path

Migration without stopping the business.

Production kept trading throughout. The new environment was built alongside it, verified against live records, then switched with a rollback ready.

  1. State 01Production system

    Analyse and architect

    Mapped checkout, payments, CRM and database dependencies before touching anything.

  2. State 02Parallel AWS environment

    Build in parallel

    Created the target environment while production remained untouched and trading.

  3. State 03Validation

    Migrate and verify

    Moved 30 GB+ of database and media while validating records against the source.

  4. State 04Cutover

    DNS switch with rollback ready

    Traffic moved to the new environment with the old one held as an immediate fallback.

  5. State 05Operational handover

    Automate and hand over

    Enabled nightly reconciliation and handed operation to the internal team.

06 / Results

The outcome was stability without rebuilding.

30 GB+Database migrated · Zero data loss

Years of transactional history preserved.

90 → 1Days to detect anomalies

Manual quarterly investigation replaced with nightly validation.

Near-zeroDowntime during migration

The business continued operating during the migration.

  • Faster storefront performance
  • Admin work isolated
  • Scheduled jobs reliable
  • Automated reconciliation
  • Immutable historical records
  • Same WordPress stack retained

07 / Technical judgement

The expensive decision was avoided.

A rebuild would have replaced working business logic.

The better solution was understanding what actually needed to change. The store, the plugins, the admin workflow and every existing order ID stayed exactly as the business already used them. Only the infrastructure underneath moved.

08 / Partner proof

“Faraz showed us the real problem was the single server, not the platform. We migrated over 30 GB with no data loss and near-zero downtime.”

CTO / Prop trading firmName withheld / NDA

Services used on this project

The engagement combined three services.

Case study · UK agency rescue

From stalled delivery to go-live in four weeks.

Read the case study

Next step

Before rebuilding your platform, understand the real bottleneck.

I help teams diagnose technical constraints, design practical solutions and deliver complex changes without unnecessary rebuilds.

Send a project brief Book a 15-min fit call

NDA-friendly · US + UK overlap · White-label by default