[RI Proposal]: Actor Based Observablity

[Andrew-UB93]

Architectural Pattern

MALT is fundamentally an actor problem.

Most observability stacks begin with centralized collection, visualization, and vendor-shaped telemetry assumptions. UBF RI should instead start from the safer primitive: isolated actors that own observable behavior and communicate through explicit, bounded messages.

MALT:

• M — Monitor: heartbeat, freshness, infrastructure/APM signals
• A — Alert: evaluate monitored facts against declared thresholds
• L — Log: persist structured facts required for alerting, audit, and telemetry
• T — Telemetry: derive metrics and actionable intelligence from observable system behavior

The framework should include:

• client libraries for safe signal emission
• polling infrastructure for systems that cannot push
• actor-owned queues/mailboxes
• idempotent polling and persistence
• support for N persistence engines
• explicit boundaries between collection, evaluation, storage, and presentation

This framework does not provide visualization. Dashboards are downstream consumers, not the core

Learning Opportunity

• Observability through MALT
• High-scale polling and ingestion
• Actor-model design at operational scale
• Safe persistence of monitoring, logging, and telemetry signals

Value Add

There is a market gap for open-source observability infrastructure that is clean, safe and efficient at scale.