Archived: Production Launch & Multi-Zone (UK/AU) Issues
Status: Archived on 2026-03-11. These issues were closed to reduce noise in the active backlog. Re-open individual issues (linked below) when the team is ready to proceed.
GitHub Issues Archived:
Part 1: Production Launch (UK South)
These three issues form the final sequence to take the PPP backend live in Azure UK South production.
Execution Order
A3 (#584) → D3 (#585) → X3 (#588)#584 — A3: Provision Azure Prod Infrastructure — UK South (ppp-uk-prod)
Labels: enhancement, infrastructure, on-hold, devops, size:xl Original issue: https://github.com/Professional-Performance-Portfolio/ppp/issues/584
Overview
Provision the production Azure infrastructure in UK South using stack ppp-uk-prod. Same resource pattern as A1/A2. DNS for api.* is not cut over here — that is a separate step (X3) after D3 is validated.
Blocked By
- #582 (A2: UAT infrastructure — prod only proceeds once UAT pattern is validated)
Blocks
- D3: Add Azure prod CD pipeline
Resources to Provision (Pulumi stack ppp-uk-prod)
Azure Resources
- [ ] Resource group
ppp-uk-prod(UK South) - [ ] PostgreSQL Flexible Server — SKU B1ms initially (can scale without Pulumi changes via portal/config)
- [ ] Azure Blob Storage — LRS Hot tier, containers:
uploads,cache - [ ] Azure Key Vault — secrets:
DATABASE-URL,STRIPE-SECRET-KEY,SENDGRID-API-KEY,GEMINI-API-KEY - [ ] Azure Container Registry (ACR) — for storing prod Docker images
- [ ] Container Apps Environment — Consumption tier
- [ ] Container App
ppp-backend-prod— image from ACR (pulled via Managed IdentityAcrPullrole), Key Vault secret refs,PORT=8080
Cloudflare DNS (api/infra-cloudflare/)
- [ ] Add staging CNAME:
api-new.professionalperformanceportfolio.com→ Container Apps FQDN (proxied) - [ ] Do NOT update
api.professionalperformanceportfolio.comyet — that is X3
Stack Config (Pulumi.ppp-uk-prod.yaml)
config:
azure:location: ukSouth
ppp:environment: prod
ppp:imageTag: latest # Overridden by CD pipeline per-deployAcceptance Criteria
- [ ]
pulumi up --stack ppp-uk-prodcompletes with no errors - [ ] Container App responds to
GET /healthz→ HTTP 200 on the staging FQDN - [ ]
https://api-new.professionalperformanceportfolio.com/healthzresolves and returns 200 - [ ] Separate PostgreSQL from UAT/dev
- [ ] Managed Identity granted
AcrPullon the prod ACR (no credentials stored in Container App)
Notes
api-new.*is the staging subdomain for prod validation before full DNS cutover- AWS Lambda continues serving
api.*in parallel until X3 is executed - B1ms PostgreSQL is sufficient initially; vertical scaling is a config change, not infra code
- This is the third slice of #192
- No
GH_TOKENor GHCR credentials needed — Managed Identity handles ACR pulls transparently
#585 — D3: Add Azure Prod CD Pipeline — trigger: merge to main
Labels: enhancement, infrastructure, on-hold, devops, ci-cd, size:l Original issue: https://github.com/Professional-Performance-Portfolio/ppp/issues/585
Overview
Add a deploy-azure-prod job to cd.yml that deploys the PPP backend to Azure Container Apps (prod) on every merge to main. Requires manual approval via GitHub environment prod before deploying.
Blocked By
- #583 (D2: UAT CD pipeline must be validated first)
- #584 (A3: Prod infrastructure must exist)
- #573 (Branch protection should be in place before prod CD is active)
Blocks
- X3: DNS cutover — prod (can only cut over once D3 is validated)
Pipeline Steps (deploy-azure-prod job in cd.yml)
- Detect changes — skip if no backend/packages changes
- Manual approval gate — GitHub environment
prodwith required reviewers configured - Log in to ACR — using ARM Service Principal (
AZURE_CLIENT_ID/AZURE_CLIENT_SECRET) - Build Docker image — tag:
${{ github.sha }} - Push to ACR
- Run Pulumi deploy —
pulumi up --stack ppp-uk-prod --non-interactive - Healthcheck —
curl --failagainst prod Container Apps FQDN (orapi-new.*until X3)
Failure Issue Lifecycle
Same deduplication pattern as D1/D2:
- Label:
cd-failure:azure-prod - Title:
CD Failure: Azure Prod Deployment - On failure → create or comment on existing issue
- On success → close open issue with resolution comment
GitHub Environment
Use environment prod with required reviewers configured (minimum 1 approver before deploy proceeds).
Labels to Create
cd-failure:azure-prod(color:#B60205)
Acceptance Criteria
- [ ] Merge to
main→ job pauses at approval gate → approved → deploys - [ ] Rejected approval → job fails cleanly (no partial deploy)
- [ ] Issue lifecycle (create / comment / close) works for
cd-failure:azure-prod - [ ] Deployed backend responds to
GET /healthz→ 200
Notes
- Required reviewers on
prodenvironment act as the production deployment gate - Until X3, the healthcheck targets
api-new.*notapi.* - Consider adding a rollback step: if healthcheck fails post-deploy, re-deploy the previous image tag
- No
GH_TOKENorpackages: writepermission needed — ACR pull is handled by the Container App's Managed Identity (AcrPullrole)
#588 — X3: Cut api (prod) DNS — Final AWS Lambda to Azure Cutover
Labels: infrastructure, on-hold, devops, size:m Original issue: https://github.com/Professional-Performance-Portfolio/ppp/issues/588
Overview
The final DNS cutover: update api.professionalperformanceportfolio.com from AWS Lambda to Azure Container Apps (prod). High-risk, high-impact — validate thoroughly with an extended smoke test before committing.
Blocked By
- #587 (X2: UAT DNS cutover must succeed and be stable for at least 24h)
- #585 (D3: Prod CD pipeline must be validated)
Tasks
- [ ] Update
api/infra-cloudflare/Pulumi:- Remove staging CNAME
api-new.*(or reassign toapi.*target) - Update
api.*CNAME from AWS Lambda URL → Azure Container Apps FQDN (output from A3 stackppp-uk-prod)
- Remove staging CNAME
- [ ] Run
pulumi upinapi/infra-cloudflare/ - [ ] Run extended smoke test (30 minutes minimum) against
https://api.professionalperformanceportfolio.com:GET /healthz→ 200- All major tRPC procedures respond correctly
- Stripe webhooks are routing correctly
- No elevated error rates in Cloudflare analytics
- [ ] Update production frontend config if needed
- [ ] Monitor closely for 48h post-cutover
- [ ] Once stable: open separate issue to decommission AWS Lambda (do not decommission immediately)
Rollback Plan
Immediate rollback (within Cloudflare):
- Revert
api.*CNAME to AWS Lambda URL - Run
pulumi upinapi/infra-cloudflare/ - AWS Lambda resumes serving traffic (kept running in parallel for this reason)
Acceptance Criteria
- [ ]
dig api.professionalperformanceportfolio.comroutes via Cloudflare proxy to Azure - [ ]
https://api.professionalperformanceportfolio.com/healthzreturns 200 from Azure prod - [ ] 30-minute extended smoke test passes with zero errors
- [ ] Stripe webhook endpoints receive and process test events correctly
- [ ] No regression in Cloudflare error rate vs pre-cutover baseline
Notes
- Do not decommission AWS Lambda in this issue. Keep it running for 2 weeks post-cutover as a rollback option.
- After X3: #192 moves into its final phase (AU East multi-region), and #194 (Grafana dashboards) is unblocked
- This completes the UK South single-region Azure deployment milestone
Part 2: Multi-Zone Infrastructure (UK & AU Data Sovereignty)
These issues cover the full Azure multi-region rollout for UK (GBP) and AU (AUD) data sovereignty compliance.
Execution Order
#192 (Azure infra) → #194 (Grafana dashboards)
#192 (Azure infra) → #1095 (country sysadmin)
#192 (Azure infra) → #636 (mobile data residency routing)#192 — Azure Multi-Region Infrastructure for UK & AU Data Sovereignty
Labels: infrastructure, architecture, on-hold, size:xl Original issue: https://github.com/Professional-Performance-Portfolio/ppp/issues/192
Sequence: 2 of 3 | Blocked by: #193 | Blocks: #194
Step Issue Status 1. Abstract AWS dependencies #193 Must complete first 2. Deploy Azure multi-region infrastructure #192 (this issue) 3. Cross-region Grafana dashboards #194 Blocked by this
Context
PPP handles sensitive medical data (surgical logbooks, CPD records, wellness assessments, doctor reviews) for UK and Australian doctors. UK GDPR and Australian Privacy Act / My Health Records Act require data sovereignty — UK data must stay in the UK, AU data must stay in Australia.
The current infrastructure is entirely AWS (Lambda, S3, Cognito, Secrets Manager) deployed via CDK. This issue tracks the full migration to Azure, managed via Pulumi, with cost minimized for pre-revenue.
Architecture reference: Multi-Region Design on the docs site.
Per-Region Azure Resources
Each region gets an identical, fully isolated resource stack:
| Resource | Tier | Est. Cost | Purpose |
|---|---|---|---|
| Container Apps | Consumption | $0-5/mo | tRPC backend (scale-to-zero, 2M req/mo free grant) |
| PostgreSQL Flexible Server | B1ms (1 vCore, 2 GiB) | ~$13/mo | Drizzle ORM — zero code changes needed |
| Blob Storage | LRS Hot | ~$1-3/mo | File storage (SAS tokens replace presigned URLs) |
| Key Vault | Standard | ~$0/mo | Database credentials, API keys, Stripe secrets |
Region total: ~$19-31/month | Two-region total: ~$38-62/month
Region Map
| Region | Azure Region | Target Doctors | Key Compliance |
|---|---|---|---|
| UK | UK South (London) | UK-registered (GMC) | UK GDPR, ISO 27001, UK G-Cloud |
| Australia | Australia East (Sydney) | AU-registered (AHPRA) | Australian Privacy Act, IRAP Protected |
Data Sovereignty Enforcement
- Separate Pulumi stacks per region — no cross-stack references
- Separate resource groups:
ppp-uk-dev,ppp-uk-prod,ppp-au-dev,ppp-au-prod - Separate databases — no replication between regions
- Separate storage accounts — LRS only (no geo-replication)
- Region-locked connections — Container Apps connect only to same-region database/storage
- Separate Key Vaults — no shared secrets across regions
Pulumi Project Structure
api/infra-azure/
Pulumi.yaml # project: embrace-azure
Pulumi.ppp-uk-dev.yaml # UK dev stack
Pulumi.ppp-uk-prod.yaml # UK prod stack
Pulumi.ppp-au-dev.yaml # AU dev stack
Pulumi.ppp-au-prod.yaml # AU prod stack
package.json
tsconfig.json
src/
index.ts # main exports
config.ts # stack config reader (region, env, tier)
resource-group.ts # resource group per region
container-app.ts # Container Apps environment + app
database.ts # PostgreSQL Flexible Server + firewall
storage.ts # Blob Storage account + containers
key-vault.ts # Key Vault + secretsStack config pattern (consistent with existing api/infra-grafana/):
# Pulumi.ppp-uk-dev.yaml
config:
azure-native:location: uksouth
ppp:environment: dev
ppp:region: ukTask Checklist
Project Setup
- [ ] Initialize
api/infra-azure/Pulumi project with@pulumi/azure-native - [ ] Create stack files for
ppp-uk-dev,ppp-uk-prod,ppp-au-dev,ppp-au-prod - [ ] Configure Pulumi local backend (consistent with
api/infra-grafana/pattern) - [ ] Add
.envtemplate for Azure credentials
UK South (ppp-uk-dev)
- [ ] Provision resource group
ppp-uk-dev - [ ] Deploy PostgreSQL Flexible Server (B1ms, SSL enforced, AES-256 at rest)
- [ ] Deploy Blob Storage account (LRS Hot) with
uploadsandcachecontainers - [ ] Deploy Key Vault with database credentials and app secrets
- [ ] Deploy Container Apps Environment (Consumption)
- [ ] Deploy Container App with PPP backend image, Key Vault secret refs, database URL
- [ ] Configure firewall rules (Container Apps → PostgreSQL, Grafana Cloud → PostgreSQL)
- [ ] Validate
pulumi previewsucceeds with zero errors - [ ] Validate deployed Container App responds to health checks
Australia East (ppp-au-dev)
- [ ] Repeat all UK South steps for
australiaeastregion - [ ] Validate complete isolation (no cross-region resource references)
Production Stacks
- [ ] Deploy
ppp-uk-prodwith same resource definitions, production-grade settings - [ ] Deploy
ppp-au-prodwith same resource definitions, production-grade settings
Verification
- [ ]
pulumi previewon each stack confirms no errors - [ ] Container Apps start and respond to health checks in both regions
- [ ] PostgreSQL accepts connections from Container Apps only (firewall validated)
- [ ] Blob Storage SAS tokens work for upload/download
- [ ] Key Vault secrets are injected into Container Apps as env vars
#194 — Cross-Region Grafana Dashboards for Super Admin Reporting
Labels: infrastructure, architecture, on-hold, size:xl Original issue: https://github.com/Professional-Performance-Portfolio/ppp/issues/194
Sequence: 3 of 3 | Blocked by: #192 | Blocks: none
Context
PPP needs cross-region reporting for super admins — showing adoption, usage, and revenue metrics across both UK and AU deployments. Grafana Cloud is already managed via Pulumi (api/infra-grafana/). This issue extends that infrastructure with regional PostgreSQL datasources and a unified dashboard, while ensuring zero PII crosses region boundaries.
Current Grafana Infrastructure
The existing api/infra-grafana/ Pulumi project manages:
- Provider: Grafana Cloud at
https://properformanceportfolio.grafana.net - Folder:
DevOps(UID:devops) - Datasource: GitHub API via Infinity plugin (for Actions billing)
- Dashboard: DevOps Overview (GitHub Actions usage metrics)
Stack: pppDev | Provider: @pulumiverse/grafana v0.16.0
Deliverables
1. Anonymized PostgreSQL Views
Create grafana_* views in each regional database containing only aggregate/anonymized metrics. No PII (names, emails, medical registration numbers, addresses) may appear in any view.
-- User adoption metrics (counts only, no PII)
CREATE VIEW grafana_adoption_metrics AS
SELECT
date_trunc('month', created_at) AS month,
country_code,
COUNT(*) AS new_users,
COUNT(*) FILTER (WHERE account_type = 'cpd') AS cpd_subscribers,
COUNT(*) FILTER (WHERE account_type = 'ppp') AS ppp_subscribers
FROM ppp_user
GROUP BY 1, 2;
-- Activity metrics (aggregate counts by type)
CREATE VIEW grafana_activity_metrics AS
SELECT
date_trunc('week', created_at) AS week,
activity_type,
COUNT(*) AS activity_count,
COUNT(DISTINCT user_id) AS active_users
FROM ppp_activity
GROUP BY 1, 2;
-- Revenue metrics (aggregate totals, no user details)
CREATE VIEW grafana_revenue_metrics AS
SELECT
date_trunc('month', created_at) AS month,
subscription_type,
COUNT(*) AS subscription_count,
SUM(amount_cents) / 100.0 AS total_revenue
FROM ppp_payment
WHERE status = 'succeeded'
GROUP BY 1, 2;
-- Wellness engagement (aggregate scores, no individual data)
CREATE VIEW grafana_wellness_metrics AS
SELECT
date_trunc('month', completed_at) AS month,
COUNT(*) AS questionnaires_completed,
AVG(total_score) AS avg_score,
COUNT(DISTINCT user_id) AS engaged_users
FROM ppp_wellness_response
GROUP BY 1;
-- Feature usage summary
CREATE VIEW grafana_feature_usage AS
SELECT
date_trunc('week', created_at) AS week,
feature_name,
COUNT(*) AS usage_count,
COUNT(DISTINCT user_id) AS unique_users
FROM ppp_feature_log
GROUP BY 1, 2;2. Read-Only Database Users
CREATE USER grafana_reader WITH PASSWORD '<from-key-vault>';
GRANT USAGE ON SCHEMA public TO grafana_reader;
GRANT SELECT ON grafana_adoption_metrics TO grafana_reader;
GRANT SELECT ON grafana_activity_metrics TO grafana_reader;
GRANT SELECT ON grafana_revenue_metrics TO grafana_reader;
GRANT SELECT ON grafana_wellness_metrics TO grafana_reader;
GRANT SELECT ON grafana_feature_usage TO grafana_reader;
-- No access to any other tables or views3. Pulumi Datasources
// api/infra-grafana/src/datasources/ppp-uk-postgres.ts
export const pppUkPostgres = new grafana.oss.DataSource(
'ppp-uk-postgres',
{
type: 'postgres',
name: 'PPP UK (PostgreSQL)',
uid: 'ppp-uk-postgres',
jsonDataEncoded: JSON.stringify({
host: process.env.PPP_UK_DB_HOST,
port: 5432,
database: process.env.PPP_UK_DB_NAME,
user: 'grafana_reader',
sslmode: 'require',
timescaledb: false,
}),
secureJsonDataEncoded: JSON.stringify({
password: process.env.PPP_UK_DB_PASSWORD,
}),
},
{ provider },
);
// ppp-au-postgres.ts follows identical pattern with PPP_AU_DB_* env vars4. PPP Metrics Dashboard (api/infra-grafana/src/dashboards/ppp-metrics.ts)
| Row | Panel | Type | Query |
|---|---|---|---|
| Adoption | UK New Users (monthly) | Time series | grafana_adoption_metrics (UK) |
| Adoption | AU New Users (monthly) | Time series | grafana_adoption_metrics (AU) |
| Adoption | Total Users | Stat | Both |
| Activity | UK CPD Activities (weekly) | Bar chart | grafana_activity_metrics (UK) |
| Activity | AU CPD Activities (weekly) | Bar chart | grafana_activity_metrics (AU) |
| Activity | Wellness Engagement | Time series | grafana_wellness_metrics |
| Revenue | UK Monthly Revenue | Time series | grafana_revenue_metrics (UK) |
| Revenue | AU Monthly Revenue | Time series | grafana_revenue_metrics (AU) |
| Revenue | Subscription Breakdown | Pie chart | grafana_revenue_metrics |
| Feature | Feature Usage Heatmap | Table | grafana_feature_usage |
5. Zero PII Guarantee
Fields that must never appear in any grafana_* view:
first_name,surname,email,phonegmc_number,ahpra_number(medical registration)address,date_of_birthpatient_name,patient_id(surgical logbook)- Any free-text fields that could contain PII
Task Checklist
Database Views
- [ ] SQL migration for
grafana_adoption_metrics - [ ] SQL migration for
grafana_activity_metrics - [ ] SQL migration for
grafana_revenue_metrics - [ ] SQL migration for
grafana_wellness_metrics - [ ] SQL migration for
grafana_feature_usage - [ ] Create
grafana_readeruser with SELECT-only grants - [ ] Deploy views to UK dev database
- [ ] Deploy views to AU dev database
- [ ] Audit: verify zero PII in all view outputs with sample queries
Pulumi Datasources
- [ ] Create
api/infra-grafana/src/datasources/ppp-uk-postgres.ts - [ ] Create
api/infra-grafana/src/datasources/ppp-au-postgres.ts - [ ] Add
PPP_UK_DB_*andPPP_AU_DB_*env vars to.envtemplate - [ ] Export datasources from
src/index.ts - [ ]
pulumi previewconfirms datasources are planned
Dashboard
- [ ] Create
api/infra-grafana/src/folders.ts— addPPPfolder - [ ] Create
api/infra-grafana/src/dashboards/ppp-metrics.ts - [ ] Build all rows/panels (adoption, activity/engagement, revenue, feature usage)
- [ ] Export dashboard from
src/index.ts - [ ]
pulumi previewconfirms dashboard is planned
Verification
- [ ]
pulumi updeploys datasources and dashboard without errors - [ ] Dashboard renders with test data from anonymized views
- [ ] PII audit: query each
grafana_*view and confirm zero PII in results - [ ] Grafana Cloud can connect to both regional databases via SSL
- [ ]
grafana_readeruser cannot access any non-grafana_*tables
#636 — Mobile: Implement Data Residency Routing — UK Users to UK Endpoint, AU Users to AU Endpoint
Labels: security, infrastructure, on-hold, mobile, size:xl Original issue: https://github.com/Professional-Performance-Portfolio/ppp/issues/636
Problem
lib/utils/trpcClient.ts reads a single TRPC_URL from Constants.expoConfig.extra.TRPC_URL. All users — UK and Australian — are routed to the same tRPC endpoint regardless of their data residency requirements.
Why It Matters
- GDPR Article 46 (UK users): Personal data must not be transferred outside the UK/EEA without adequate safeguards
- Australian Privacy Act 1988, APP 8: Cross-border disclosure rules apply to health data
- Mixing UK and AU medical records in the same data store and region violates both regulations
- Regulators (ICO for UK, OAIC for AU) can issue significant fines for data residency violations
Acceptance Criteria
- [ ] User's country/region is available in the JWT claims or
userprofile - [ ]
trpcClient.tsselects endpoint based on user region - [ ] UK users:
TRPC_URL_UK(EU/UK region endpoint) - [ ] AU users:
TRPC_URL_AU(AP-Southeast-2 endpoint) - [ ] EAS
app.config.jsincludes bothTRPC_URL_UKandTRPC_URL_AUvariables - [ ] Integration test verifies correct endpoint selection per region
#1095 — feat(admin): Country-Level Sysadmin Role and Instance-Scoped Reference Data Management
Labels: enhancement, backend, on-hold, size:l Original issue: https://github.com/Professional-Performance-Portfolio/ppp/issues/1095
Problem
PPP will run as two separate instances in two locations for data sovereignty:
- UK instance (Azure UK South) — UK-registered doctors (GMC/ISCP)
- AU instance (Azure Australia East) — AU-registered doctors (AHPRA/RACS)
A country-level sysadmin role is needed — someone responsible for managing their country's instance reference data (hospitals, SNOMED→WBA mappings, training body configuration) without any access to the other country's data.
Architecture: Two Instances, Not Multi-Tenant
UK Instance (Azure UK South) AU Instance (Azure Australia East)
├── pppUkUat / pppUkProd ├── pppAuUat / pppAuProd
├── Own PostgreSQL database ├── Own PostgreSQL database
├── Own Container App ├── Own Container App
├── Own Key Vault ├── Own Key Vault
├── UK hospitals (NHS) ├── AU hospitals
├── UK SNOMED→WBA mappings (ISCP) ├── AU SNOMED→WBA mappings (RACS)
├── GMC regulatory validation ├── AHPRA regulatory validation
└── UK sysadmin manages this └── AU sysadmin manages thisCountry-Level Sysadmin Capabilities
Can do (within their instance only):
- Hospital management — Import, add, edit, deactivate hospitals for their country (#267)
- SNOMED→WBA mapping curation — Review and confirm SNOMED concept → WBA type mappings (#134)
- Training body configuration — WBA types, CPD categories, and assessment frameworks
- Regulatory body settings — GMC validation rules (UK) or AHPRA validation rules (AU)
- User administration — Existing
isAdmincapabilities - Reference data import — Bulk import from
ppp-data-lookupPoC data files
Cannot do:
- Access the other country's instance (enforced by infrastructure)
- Modify system configuration (env vars, feature flags, infrastructure) — DevOps only
- Access Grafana cross-region dashboards (#194) — super user only
Implementation Approach
The existing isAdmin flag already scopes to the instance since each country is a separate database. What is needed:
- Admin screens for reference data — new Settings tabs for hospitals, SNOMED→WBA mappings, training body config
- Bulk import functionality — import reference data from
ppp-data-lookupPoC CSV/JSON files - Instance-aware admin UI — show country context (e.g., "UK Hospitals", "RACS Assessment Types") based on
INSTANCE_COUNTRYenv var - Instance country config — add
INSTANCE_COUNTRYenv var (or derive fromINSTANCE_ID)
Mobile App: Instance Selection
The mobile app (#636) needs an initial instance selector:
- User chooses "United Kingdom" or "Australia" at first launch
- Selection determines which tRPC endpoint (UK or AU) the app connects to
- Selection is locked — user cannot switch without creating a new account
Acceptance Criteria
- [ ] Each instance has a country identifier (env var or config) that drives admin screen context
- [ ] Admin screens display country-appropriate labels ("UK Hospitals" vs "AU Hospitals")
- [ ] Hospital management admin screen with bulk import from CSV (#267)
- [ ] SNOMED→WBA mapping curation screen with training-body-appropriate WBA types (#134)
- [ ] Training body configuration screen (ISCP framework for UK, RACS framework for AU)
- [ ] No cross-instance data access possible (verified by test)
- [ ] Reference data import from
ppp-data-lookupPoC works for both UK and AU datasets
Dependencies
- Blocked by: #192 (Azure multi-region infrastructure — instances must exist first)
- Related: #134 (SNOMED→WBA mapping — needs country/training-body dimension)
- Related: #267 (hospital import — needs to be instance-scoped)
- Related: #636 (mobile data residency routing — needs instance selector UX)
- Related: #1087 (SNOMED integration epic)
- Related: #194 (Grafana dashboards — super user cross-region reporting)
- Data source:
ppp-data-lookupPoC → imported per-instance via admin screens