Infrastructure Prerequisites

Infrastructure Prerequisites for Capacity Private Cloud

To ensure the seamless performance of the Capacity Private Cloud (formerly LumenVox) containerized architecture, clients are responsible for providing and supporting several external software components.

These "Best-in-Class" solutions handle internal messaging, caching, and data persistence. To maintain high availability (HA), scalability, and robust disaster recovery, we recommend deploying distributed clusters for each component in production environments rather than single instances.

Core Messaging & Caching

These components facilitate the high-speed communication and temporary data storage required for real-time speech processing.

RabbitMQ | Message Queuing

Used for all internal messaging between microservices, including transaction queuing, audits, and management information. Once a microservice processes a request, it is cleared from the queue.

• Deployment: Independent client/partner implementation.
• Version: See Support Matrix.

Redis | In-Memory Cache

Acts as the primary database cache, holding session, audio, and interaction IDs. Redis is the engine that receives and processes audio for speech processing and caches transaction info across the architecture.

Version: See Support Matrix.

Data Persistence (Databases)

The Capacity Private Cloud utilizes two distinct database types to manage binary data and relational metadata.

MongoDB | Document Database

A key-value, document-oriented database used to store audio, grammars, and SSML files in binary format. This is critical for audit trails, troubleshooting, and performance refinement via the Analysis Portal.

• Customization: Clients may customize the MongoDB database name (excluding KubeAdm deployments).

• Storage Metrics:

  • TTS: 2.5kb per minute

  • ASR & Transcription: 500kb per minute

Version: See Support Matrix.

PostgreSQL | Relational Database

A SQL-compliant repository for persistent metadata, licensing, and reporting data. This handles all transactional data storage.

• Customization: A single database is used; the name can be customized (excluding KubeAdm deployments).

• Storage Metrics:

  • TTS: 2.5kb per minute

  • ASR & Transcription: 25kb per minute

Version: See Support Matrix.

Persistent Storage

In addition to the databases above, the environment requires Persistent Storage (SSD recommended) for system-critical assets:

• Standard Provisioning: 100GB SSD (Minimum recommendation; actual sizing depends on volume).
• Usage: Used for storing ASR/TTS models, Voice Biometrics, and Grammar files.
• Scalability: Storage needs will increase as additional language models or extensive grammar caches are implemented.