Skip to main content

Developer Platform (PaaS)

The Okustera Developer Platform provides serverless runtimes, an enterprise API gateway, message queuing, workflow orchestration, universal artifact registries, centralized observability, and AI/ML inference infrastructure designed to eliminate operational friction and accelerate application delivery.

Developer Platform Services​

API Gateway & Ingress (/developer/gateway)​

High-performance dynamic ingress and API gateway powered by Apache APISIX:

  • Sub-millisecond proxy routing, automated Let's Encrypt SSL/TLS certificate issuance.
  • Weighted Canary Traffic-Split releases, blue/green routing, and header-based traffic shifting.
  • Enterprise security: WAF (OWASP CRS), rate limiting, forward-auth token validation, and IP whitelisting.
  • Specialized AI Gateway: unbuffered SSE streaming, model payload routing, Valkey semantic caching, and token FinOps extraction.
  • Declarative management via Terraform (okustera_apisix_route) and Kubernetes CRDs (ApisixRoute).

Serverless Functions (/developer/serverless)​

Event-driven function compute powered by OpenFaaS with Google gVisor kernel isolation:

  • Write code in Python, Node.js, Go, Rust, or custom Dockerfiles.
  • Instant cold-starts with strict multi-tenant kernel sandboxing (runsc).
  • Zero-trust agentic code execution sandboxes with sub-50ms ephemeral container lifecycles.
  • Declarative scaling to zero and automatic concurrency bursting.
  • Managed via Terraform (okustera_function) or the Okustera Cloud Portal Web IDE.

Message Queuing (/developer/messaging)​

Enterprise distributed message brokering and asynchronous task processing powered by RabbitMQ Cluster Operator:

  • High-throughput AMQP 0-9-1, AMQP 1.0, MQTT, and STOMP message protocols.
  • Raft-based Quorum Queues for robust message durability and partition tolerance.
  • Built-in web management console, dynamic virtual hosts, and TLS encryption.
  • Native autoscaling of consumer workloads via KEDA.

Workflow & Data Orchestration (/developer/workflows)​

Enterprise-grade data pipeline scheduling and workflow automation powered by Managed Apache Airflow:

  • Zero-license-tax open-source alternative to AWS MWAA and GCP Cloud Composer.
  • Serverless task elasticity using KubernetesExecutor on tenant clusters.
  • Native backing with CloudNativePG PostgreSQL 16 metadata and Ceph RGW S3 DAG/log storage.
  • Automated S3 synchronization and drag-and-drop Cloud Portal UI DAG upload.
  • Declarative management via Terraform (okustera_workflow_dag, okustera_workflow_status).

Universal Artifact Registry (/developer/registry)​

Centralized sovereign package registry powered by Artifact Keeper:

  • OCI-compliant container registry with automated vulnerability scanning (Trivy).
  • Multi-format repository support: Helm charts, PyPI packages, npm modules, and Maven artifacts.
  • OCI Model Packages and LoRA adapters using ORAS / ModelKit integrated with the Model Foundry.
  • Integrated IAM role-based token authentication and private tenant namespaces.

Observability & Telemetry (/developer/observability)​

Full-stack operational visibility and proactive telemetry:

  • Dynamic Service-Aware Dashboards: Auto-synthesized Grafana dashboards dynamically provisioned for active functions, databases, compute VMs, and workflows with subtractive pruning.
  • Curated Platform Administration: Cluster-wide control plane telemetry for OpenStack, Apache APISIX Ingress, OpenFaaS Serverless, and Managed Airflow.
  • Centralized Structured Logging: High-throughput log streaming and LogQL querying with Grafana Loki.
  • Generative AI Tracing: Real-time LLM distributed tracing, TTFT latency profiling, and token rating with Langfuse.
  • Autonomous Predictive Detection: Continuous proactive anomaly forecasting and health auditing via omc-sentinel.

FinOps, Metering & Billing (/developer/billing)​

Transparent real-time infrastructure cost tracking and budget governance:

  • Real-time Month-to-Date spend metering and forecast projection across compute, storage, and databases.
  • AI token rating (prompt and completion tokens) and real-time FinOps accounting.
  • Customizable monthly budget caps and threshold alerts with automatic quota protection.
  • Secure payment method tokenization and instant Stripe checkout session funding.

AI Inference PaaS & Model Foundry (/developer/ai-ml) (Roadmap Preview)​

Accelerated compute and sovereign foundation model platform:

  • The OMC Model Foundry: Curated foundation blueprints (DeepSeek-R1, Llama 3.3, Qwen 2.5, StarCoder2), automated SafeTensors security audit, and Hugging Face mirror.
  • High-Throughput Serving & Dynamic Multi-LoRA: vLLM via KubeRay RayService, PagedAttention, and hot-swapping up to 64 custom tenant models on 1 base GPU in under 10ms.
  • Asynchronous Frontier MoE Batch Tier: Kubernetes Kueue and Colibrì streaming sparse experts from NVMe for 284B–744B models.
  • Vector DBaaS & RAG: Dedicated Qdrant cluster with scalar quantization and hybrid search.
  • Decoupled Multi-Node Scaling: Decoupled AI compute nodes with NVIDIA RTX 4090, L40S, or A100/H100 GPUs and MIG partitioning.