k8s-cluster

Computation Resources(actualy)

Tipo	Processors	CPU Cores	RAM	Total
Servers(12x)	Intel Xeon(amd64)	12-80	16-192GB	732GB
Computer(18x)	Intel i5/i7 (amd64)	4-16	4-16GB	180GB
Edge(6x)	Rasberry Pi 4(amd64)	4	8GB	46GB

Computation Resources(expansion plan)

1. Edge Ai

• 8 x NVIDIA Jetson Orin™ Nano 8GB (40 TOPS)

2. Setup 5G (Software Defined Radio USRP)

• 4x USRP B210
• 2x Dell XPS - Intel i7 16GB
• 2x USRP B200-mini

3. Switches OpenFlow nativos

• 1x Centec v350
• 48 P 1Gbps + 4 P 10Gbps
• 2x Centec v330
• 48 P 1Gbps + 4 P 10Gbps

Control and Orchestration Software

Aplications Deploy

Deploy use Helm Charts (“packets” manager similar to apt/yum). For exemple:

$ helm repo add dsri https://maastrichtu-ids.github.io/dsri-helm-charts/
$ helm repo update
$ helm install jupyterlab dsri/jupyterlab

The services and deploy aplications use YAML.

Container Network Interface (CNI)

Network Service

• IPs management (clusterIP)
• Ingress creation (for external traffic)
• LoadBalancers

Basead in eBPF

• Perfomance

• flexbility

  Observability with Hubble

Container Storage Interface (CSI)

services of Storage

• PersistentVolumes through PersistentVolumeClaims (PVC)
• Storage per Pod(block)
• Storage shared (file System)

Total Storage: 16.8GB

• SDD: 10.9 TB
• HDD: 5.9TB

Monitoring

Prometheus

• Time Series Database
• Data colleted through exporters
• Queries via PromQL

Grafana

• Information view
• Multiple sources(including Prometheus)

Loki

• Distributed log manager

  

Service Mesh Observability

High level service network Service Mesh Sidecars

• Envoy proxy injection
• Observabilidade, tracing, debugging

Services

• Gateways
• Virtual services
• Extern acess

Security

• Acess policies
• Encryption

Storage as a Service (SaaS)

Dropbox-style storage solution

• Multiples users
• Multiples devices
• Acess control
• Versioning
• Sharing

Serverless Computing

• Event Driven Applications (Cloud Events)
• Message Routing
• Auto Scaling