Oriental Nippon Development
Back to portfolio
Case studyTelecoms group of eleven legal entitiesInitiating: not yet delivered

Planned: A Zero-Weekend ERP Upgrade for an 11-Entity Telecoms Group

Eleven entities, one ERP database three major versions behind, five bespoke modules and a million-plus journal lines. This programme is scoped and priced. Here's how we plan to move it entity by entity with rollback as a route change.

Industry
Telecommunications
Region
Global
Engagement
Lab-type ODC (dedicated team)

Outcomes

  • Target: minutes of downtime per entity
  • Rollback = route change
  • Documented, tested recovery

Stack

Odoo · OpenUpgrade · PostgreSQL · Reverse proxy · n8n · Meltano · GitHub Actions

Written byOriental Nippon Team

This engagement is being initiated. What follows is the plan we proposed, not completed work.

A telecoms group of eleven legal entities runs everything on one shared, heavily customised ERP database, three major versions behind current. Around forty-seven people use it daily across accounting, purchasing, HR, helpdesk and a hospitality business owned by one entity.

The usual answer is a big-bang upgrade weekend. We proposed the opposite: move one company at a time, with the old system untouched until the last wave is proven.

What we found

  • Deep customisation. Low-code changes at core level, including ledger balance behaviour and document rendering, plus five bespoke modules and five live integrations.
  • A heavy database. Over a million journal lines, and one stored-signature table taking up roughly half the database.
  • Fragile automation. A scheduled helpdesk routine referenced fixed company IDs and custom fields, against an interface that has changed twice since.
  • Weak recovery. Database backups ran, but the file store had no schedule and there was no written recovery plan.
  • Outdated servers. The server stack was below the target version's minimum.
Figure
Legal entities11
Daily users~47
Versions behind3 major releases
Bespoke modules / live integrations5 / 5
Journal lines1,000,000+
Indicative programme length4–6 months

What we proposed

  • A parallel system

    A new, correctly sized instance beside the live one, on current OS, database and runtime.

  • A routing layer

    A reverse proxy sends each company's users to the old or new system, so rollback is just a route change.

  • One-way data movement

    Reference data via lightweight ETL; transactional data moved once per company at switch-over.

  • Rebuilt customisation

    Bespoke modules, core changes, document layouts and scheduled routines rewritten properly.

  • A cleanup pass

    Binary data moved to the file store, closed fiscal years archived, dormant users and unused modules removed, and a backup schedule and recovery plan written.

Each of the eleven companies moves through the same states on its own date:

Technology and the role it played

TechnologyRole in this project
OdooCurrent-version target system
OpenUpgradeVersion-to-version database migration
PostgreSQLSlimmed, migrated data
Reverse proxyRoutes each company's users to the old or new system
n8n and MeltanoReference-data movement and integrations
GitHub ActionsPipelines for three environments

How we would deliver it

The pilot wave moves HR for the smallest entity, proving the full mechanism where failure costs least. The two largest entities and accounting go last, timed to a fiscal-period boundary.

  1. Discovery. Separate real requirements from inherited habits, and review every custom module, report and job.
  2. Foundation. Three environments, source control, pipelines, the routing layer and the cleanup.
  3. Pilot wave. Smallest entity, lowest-risk module.
  4. As-is waves. Company by company, lowest risk first.
  5. Redesign and stabilise. Process improvements, hardening and retirement of the old system only once every wave is proven.
Delivery timelineIndicative total: 14–25 weeks
  1. 01Discovery

    1–2 wks

    Scope, wave order, access, risk grading

  2. 02Environments and cleanup

    1–2 wks

    Three environments, CI/CD, routing layer, database slimming

  3. 03Pilot wave

    2–3 wks

    Smallest entity, lowest-risk module, full round trip proven

  4. 04As-is migration waves

    6–10 wks

    Entity by entity; custom modules, reports and integrations rebuilt

  5. 05Redesign and stabilisation

    4–8 wks

    Process redesign, training, hardening, hand-over

Risks and how we handled them

RiskHow we handled it
A failed big-bang weekendCompany-by-company waves, with the old system unchanged
No way back after switch-overRollback is a routing change, tested per wave
Core customisations breakingRebuilt as proper modules with an upgrade path
Data loss during migrationFile-store backup schedule and written recovery plan added first

What it would change

Downtime measured in minutes per company instead of a weekend for the group, a supported platform, customisations rebuilt as proper modules with an upgrade path, and a tested recovery position where there was none.

We would rather state that plainly than promise a percentage. This programme removes a category of risk. It does not, by itself, make the business faster.

What the client would keep

  • A migration runbook and per-wave rollback procedures
  • Reconciliation reports
  • Rebuilt custom modules, in their own repository
  • Integration documentation
  • The pre-migration audit, as a permanent record of what the system contained

Sources and further reading

  1. OpenUpgrade, OCA on GitHub
  2. Strangler fig application, Martin Fowler
  3. Backup and restore, PostgreSQL documentation

Facing something similar? This is the kind of work our Enterprise & Digital Transformation team delivers every day. Talk to us about your stack.