Looking for the distributed, multi-node mesh architecture?
Check out FlyOps-2.0 — a resilient 19-neuron connectome cluster featuring dynamic L7 failover, CoreDNS discovery, and Grafana Node Graph visualization.

English | Русский

Drosophila Connectome Simulator & Observability Stack

A cloud-native demo project: a Python microservice simulating spike propagation dynamics across a biological connectome graph of Drosophila melanogaster. The service is containerized, deployed to a local Kubernetes cluster (Minikube), and monitored using the Prometheus Operator and Grafana.

Overview

• Microservice (Python): Simulates neural spike dynamics across biological graph layers (ORN, PN, KC, MBON, DN), calculates synapse latency, and exposes OpenMetrics via prometheus_client.
• Containerization & Orchestration: Packaged via Docker, deployed to Minikube using declarative manifests (Deployment, ClusterIP Service).
• Monitoring (Prometheus Operator): Target discovery and metric collection automated via Kubernetes Custom Resource Definition (ServiceMonitor).
• Visualization (Grafana & PromQL): Dashboards configured to track spike rates, p95 latency, and active neuron clusters.

Architecture

[ Synapses CSV ] ──> [ Python Simulator Pod ] ──(Exposes :8000/metrics)
                             ▲
                             │ Scrapes every 5s
                    [ ServiceMonitor (CRD) ]
                             │
                  [ Prometheus Operator ] ──> [ Grafana Dashboard ]

Graph Model & Tech Stack

The simulation parses synapses.csv across functional biological layers:

• ORN: Olfactory Receptor Neurons (Sensory input)
• PN: Projection Neurons
• KC: Kenyon Cells (Pattern recognition)
• MBON: Mushroom Body Output Neurons
• DN: Descending Neurons (Motor commands)

Core Stack: Python 3.11, Docker, Minikube (Kubernetes), CoreOS Prometheus Operator, Grafana.