PostQS/Platform/Architecture

A unified architecture for modern data platforms.

Storage, execution, identity, and security must be designed together as one system. Rather than assembling independent components, PostQS products are built around a cohesive architecture where each layer reinforces the others.

Architectural goals

Designed to behave consistently - across loads, environments, and time.

These goals guide the structure of every platform component.

  • Support continuous data access
  • Operate on encrypted data representations
  • Enforce identity and access natively
  • Maintain persistent state across memory and storage
  • Remain predictable under real-world workloads
PostQS architecture diagram.
High-level structure

Components that operate independently - and as one system.

Each component is designed to operate independently, while still participating in a unified execution and trust model.

Enclave Data Platform

Execution and storage. Encrypted, in-memory-first, deterministic.

EnclaveVault

Identity, authorization, and cryptographic control. A native trust boundary.

Encrypted persistence

Coordinated encrypted memory, write-ahead logs, and blob storage.

Standard interfaces

SQL via PostgreSQL wire protocol, Apache Arrow analytics, admin APIs.

Secure inter-component comms

Shared-memory transport on a node, encrypted RPC across nodes.

TRNG hardware entropy

A platform-level entropy resource for cryptographic operations.

Execution & storage

In-memory-first SQL on encrypted, record-referenced data.

Execution stages operate on record references and metadata. Intermediate results are minimized. Access checks are evaluated as part of execution.

  • Relational SQL execution with standard semantics
  • Persistent encrypted state - memory, WAL, blob storage
  • Index-free, deterministic record references
  • Predictable behavior across restarts and recoveries
Explore EDP
CLIENT SQL · Arrow · Admin API EXECUTION Parse → AuthZ → Plan → Execute → Result Operates on record references; minimized intermediates PERSISTENCE Encrypted Memory Encrypted WAL Encrypted Blob Store TRUST EnclaveVault Identities · authZ tokens · key lifecycle TRNG-backed entropy Same trust model across components
Interfaces & deployment

Standard access on top. Purpose-built behavior underneath.

EDP exposes familiar interfaces while preserving a distinct internal execution and security model. The architecture supports a range of deployment patterns without altering the core model.

Drop-in PostgreSQL wire

Bring an existing Postgres app - standard clients and drivers connect without code changes.

Native direct protocol

For new applications - skip the emulation layer for materially better performance.

Apache Arrow API

Columnar interchange for analytics and pipelines, alongside SQL and native access.

Single-node systems

Run the full platform on one node with shared-memory transport.

Multi-container & clusters

Distributed deployments use encrypted RPC and the same trust model.

On-prem, hybrid, controlled

Choice of environment doesn't alter execution, identity, or persistence.

One unified platform.

Execution, storage, identity, and security are not optional features or add-ons. They are intrinsic to how the platform operates.