HMP-0005_06_10

Источник: HMP-0005_06_10.md

HyperCortex Mesh Protocol (HMP 5.0.8) - модульное представление

• Полный документ: HMP-0005.md
• Список разделов: HMP-0005_index.md

---

6. Core protocols

Смотреть 6. Core protocols - общая часть

---

6.10 Distributed Container Propagation

6.10.1 Purpose

The distributed container propagation mechanism defines how containers are stored, replicated, and exchanged between agents within the HMP network.

While MCE provides direct container exchange and MRD ensures targeted delivery to specific recipients,
this section describes the principles of resilient and coherent data propagation across multiple participants,
including replication, filtering, and index synchronization.

---

6.10.2 Core Principles

1. Self-propagation
   Each agent autonomously decides which containers to retain, forward, or index.
   These decisions depend on the agent’s declared roles, trust level (RTE), interests, and local policies.

2. Semantic Selectivity
   Containers are propagated based on their semantic context (class, tags, and other metadata) rather than randomly.
   This prevents network overload and unnecessary duplication.

3. Containers with Limited or Unlimited Lifetime
   Depending on purpose:

   • ephemeral containers (e.g., goal, task, vote) SHOULD have a short ttl;
   • long-lived containers (e.g., document, research, artwork — potential future classes) MAY have a high ttl or no expiration at all.

4. Adaptive Replication
   The number of container copies within the Mesh is determined dynamically,
   according to trust relationships, topic relevance, and peer activity.

5. Index Merging (container_index merge)
   Agents periodically reconcile and merge their indices, unifying and refreshing records of available containers.
   This maintains global consistency while minimizing transport overhead.

---

6.10.3 Propagation Types

Type	Description
Full Sync	Complete exchange of container_index datasets between trusted peers. Commonly used between archival or mirror nodes.
Selective Sync	Propagation filtered by topics, roles, trust levels, and container lifetime. This is the most typical mode.

---

6.10.4 Propagation Coordination

Container propagation is coordinated using existing MCE structures:

Container	Purpose
container_index	Stores and publishes the list of available containers.
container_delta	Transfers incremental index updates (additions, modifieds or removals).
container_request	Requests specific containers from peers.
container_response	Returns requested containers or their metadata.
container_ack	Confirms receipt of a container.

When it is necessary to log the fact of a container batch propagation, this MAY be represented as a workflow_entry with the subclass "propagation_log" or "index_sync".

---

6.10.5 Integration with Other Layers

Layer	Role
MCE (5)	Provides low-level container and index exchange.
MRD (6.8)	Handles routing and delivery between agents.
RTE (6.9)	Determines replication priorities and trust-based filters.
SAP (6.7)	Enables recovery of historical or missing containers.

---

6.10.6 Security and Load Management

• Rate and size limiting: agents regulate transfer volume based on trust and network quotas.
• Interest filtering: containers are exchanged only between agents for whom they are relevant.
• Source verification: only containers with valid signatures and trusted provenance are accepted.
• Private propagation: encrypted exchange is permitted between trusted peers when confidentiality is required.

---

Summary

The container propagation system in HMP represents a self-organizing information flow mechanism, where replication, filtering, and retention depend on trust, interest, and contextual relevance.

Through these principles, the Mesh remains resilient, balanced, and cognitively guided — operating without centralized servers or fixed delivery routes.