Architecture Overview

On-Premise Agent Architecture

The EPMware On-Premise Agent provides a secure bridge between EPMware Cloud applications and your on-premise target systems. This architecture ensures that your internal applications remain protected while enabling seamless metadata management.

EPMware On-Premise Agent Architecture - Showing the flow between EPMware Cloud, On-Premise Agent, and Target Applications

Key Components

1. EPMware Cloud Application

The cloud-based EPMware application serves as the central hub for:

Metadata request management
Workflow orchestration
User interface and dashboards
REST API endpoints
Deployment scheduling

2. On-Premise Agent

The agent acts as an intermediary, providing:

Secure Communication - REST API-based authentication with tokens
Command Execution - Processes deployment commands from EPMware
Status Reporting - Returns execution results to EPMware
Polling Mechanism - Checks for pending tasks at configured intervals

3. Target Applications

Supported on-premise applications include:

Hyperion Financial Management (HFM)
Hyperion Planning
Essbase
Other EPM applications

Communication Flow

1. Polling Process

sequenceDiagram
    participant Agent
    participant EPMware
    participant Target

    Agent->>EPMware: Poll for tasks (every 30 seconds)
    EPMware-->>Agent: Return pending tasks
    Agent->>Target: Execute deployment
    Target-->>Agent: Return results
    Agent->>EPMware: Report status

2. Data Flow

Request Initiation - Users submit metadata requests through EPMware UI
Workflow Processing - Requests move through approval workflow
Deployment Queue - Approved requests are queued for deployment
Agent Polling - Agent checks for pending deployments
Command Execution - Agent executes deployment to target application
Status Update - Results are reported back to EPMware

Security Architecture

Authentication Layers

EPMware Authentication
REST API token-based authentication
36-character alphanumeric tokens
Secure HTTPS communication
Target Application Authentication
Application-specific credentials
Encrypted password files
Service account authentication

Network Security

Firewall Protection - Only outbound connections required
SSL/TLS Encryption - All communication encrypted
No Inbound Ports - Agent initiates all connections
Token Rotation - Periodic token refresh capability

Deployment Models

Single Agent Configuration

Best for organizations with: - Single data center - Co-located EPM applications - Centralized IT management

Single agent serving multiple target applications

Multiple Agent Configuration

Recommended for: - Geographically distributed applications - Different security zones - Load distribution requirements

Multiple agents for distributed deployment

Hybrid Cloud Configuration

For organizations using both cloud and on-premise applications:

Hybrid architecture with both cloud and on-premise targets

Agent Components

Core Files

File/Directory	Purpose
`epmware-agent.jar`	Main agent executable
`agent.properties`	Configuration settings
`ew_target_service.sh`	Service startup script
`logs/`	Agent log files
`temp/`	Temporary working directory

Process Architecture

The agent runs as: - Windows: Scheduled Task or Windows Service - Linux: Background daemon process

Performance Considerations

Polling Frequency

Default: 30 seconds
Minimum: 10 seconds (high-frequency deployments)
Maximum: 300 seconds (low-frequency deployments)

Resource Usage

Memory: ~512MB - 2GB depending on deployment size
CPU: Minimal during polling, increases during deployment
Network: Burst traffic during metadata transfer
Disk I/O: Temporary files during processing

High Availability Options

Active-Passive Configuration

Primary agent handles all requests
Standby agent activated on failure
Manual or automated failover

Load Balanced Configuration

Multiple agents with shared workload
Application-based distribution
Improved throughput for large deployments

Monitoring and Management

Health Checks

The agent provides several monitoring capabilities:

Connection status verification
Deployment success/failure metrics
Response time tracking
Error logging and alerting

Log Management

Two primary log files: 1. agent.log - Deployment activities and errors 2. agent-poll.log - Polling activity tracking

Integration Points

EPMware Integration

REST API for communication
JSON-based message format
Asynchronous task processing

Application Integration

HFM: HFM utilities and command-line tools
Planning: Planning utilities and encrypted credentials
Cloud EPM: EPM Automate utility
Custom: Scriptable integration points

Scalability Considerations

The architecture supports:

Vertical Scaling: Increase server resources for larger deployments
Horizontal Scaling: Add agents for distributed processing
Queue Management: Handle multiple concurrent deployments
Batch Processing: Group deployments for efficiency

Best Practice

For production environments, implement agent redundancy to ensure continuous availability of metadata deployment capabilities.

Cloud vs On-Premise

While EPMware can be deployed either on-premise or in the cloud, the agent architecture remains consistent, with only the connection endpoints changing.

Next Steps

Now that you understand the architecture: