🏗️ Platform Architecture Overview
Hub, PM4ML, and Tazama (Cloud & On-Premise)
This document describes the reference architecture for deploying the Mojaloop Hub, PM4ML (Payment Manager for Mojaloop), and Tazama platforms across cloud and on-premise datacenter environments.
The architecture focuses on deployment topology, platform responsibilities, security boundaries, and operational separation, without exposing environment-specific or sensitive implementation details.
Note
This document presents a reference architecture based on commonly adopted deployment patterns.
Specific implementations may vary depending on regulatory, organizational, and operational requirements.
1. Architectural Goals
The architecture is designed to:
- Support secure financial transactions
- Enable cloud, on-premise, and hybrid deployments
- Provide operational resilience using a Primary–Standby model
- Maintain clear trust and responsibility boundaries between platforms
- Scale predictably based on transaction throughput
2. Deployment Topology (Primary–Standby)
The platform is deployed across two independent sites:
Primary Site
- Active production environment
- Handles live transaction traffic
- Deployed in:
- Public cloud or
- On-premise datacenter
Standby Site
- Secondary environment with reduced capacity
- Maintained for service continuity
- Deployed in a separate cloud region or physical datacenter
┌───────────────┐ Secure Connectivity ┌───────────────┐
│ Primary Site │ ◀────────────────────────────▶ │ Standby Site │
│ │ │ │
│ Hub │ │ Hub │
│ PM4ML │ │ PM4ML │
│ Tazama │ │ Tazama │
└───────────────┘ └───────────────┘
Both sites share the same logical architecture. Differences are limited to capacity sizing, not functionality.
3. Platform Components
3.1 Mojaloop Hub
- Central transaction switching platform
- Enables interoperability between participants
- Deployed as containerized services
- Acts as the trust authority for DFSP onboarding
3.2 PM4ML (Payment Manager for Mojaloop)
- DFSP-facing payment management platform
-
Can be deployed:
-
Per DFSP, or
- As a shared service
- Supports both cloud-hosted and on-premise environments
- Establishes business trust with the Hub
3.3 Tazama
- Transaction monitoring and fraud detection platform
- Consumes transaction events for analysis
- Can be deployed centrally or regionally
- Operates independently of transaction processing paths
4. Infrastructure Architecture
Runtime Platform
- Kubernetes used as the common orchestration layer
- Consistent deployment patterns across environments
Networking
- Private, encrypted communication between platforms
- Hybrid connectivity supported for cloud ↔ on-premise scenarios
Traffic Management
- Controlled ingress and egress for platform services
- Clear separation between internal and external traffic paths
Data Services
- Databases and messaging systems sized per expected throughput
- Logical isolation maintained between platforms
5. Security Architecture (High Level)
┌────────┐ mTLS + JWS ┌─────────┐
│ Hub │ ◀────────────────▶ │ PM4ML │
└────────┘ └─────────┘
│
│ Transaction Events
▼
┌─────────┐
│ Tazama │
└─────────┘
Security Principles
- Encrypted communication for all inter-platform traffic
- Strong platform identity and authentication
- Separation of infrastructure trust and business trust
- Centralized secrets and key management
Detailed mTLS and JWS mechanisms are documented separately.
6. Availability and Resilience
- High availability within each site
- No active-active transaction processing across sites
- Standby site maintained in a ready state
- Recovery and failover procedures defined outside this document
7. Operational Model
- Infrastructure provisioned using automation
- Platform deployments follow GitOps practices
- Centralized monitoring, logging, and alerting
- Consistent operations across cloud and on-premise environments
8. Scope Clarification
This document covers:
- Platform roles and responsibilities
- Deployment topology
- High-level security and connectivity patterns
This document does not include:
- Environment-specific configurations
- Backup and restore procedures
- Incident response or failover runbooks
- Regulatory or compliance controls
9. Key Takeaways
- A single architecture supports cloud, on-premise, and hybrid deployments
- Primary–Standby model balances resilience and cost
- Hub, PM4ML, and Tazama have clearly separated responsibilities
- Security is enforced at both transport and application layers
10. Summary
The Mojaloop Hub, PM4ML, and Tazama platforms are deployed using a consistent, secure, and flexible architecture that supports cloud and on-premise environments while maintaining clear trust boundaries and operational resilience.