|
| 1 | +# Architecture |
| 2 | + |
| 3 | +Un schéma d'implémentation détaillé sera établi pour décrire l'architecture globale de l'infrastructure, y compris les différents composants, leur interaction et leur déploiement. |
| 4 | + |
| 5 | + |
| 6 | +## Application |
| 7 | + |
| 8 | + |
| 9 | +### Proxy: Traefik |
| 10 | + |
| 11 | +* Application router |
| 12 | +* High-availability by fault tolerance |
| 13 | +* Load-balancing by requests distribution |
| 14 | + ALB ? |
| 15 | +* Helm chart |
| 16 | +* Version |
| 17 | + |
| 18 | + |
| 19 | +### Application: FastAPI |
| 20 | + |
| 21 | +* Python |
| 22 | + * Version: |
| 23 | + * Packaging |
| 24 | +* Project management? Poetry? |
| 25 | + |
| 26 | + |
| 27 | +### Databases: PostgreSQL |
| 28 | + |
| 29 | +* Application is using the well known and battle-tested OSS PostgreSQL database management system. |
| 30 | + |
| 31 | + FIXME pg version |
| 32 | + |
| 33 | +* Use a specialized Kubernetes operator to manage the PostgreSQL cluster |
| 34 | + |
| 35 | + Zalando's **postgres-operator** is chosen because... |
| 36 | + FIXME main reason(s) to choose it over alternatives |
| 37 | + |
| 38 | + Alternative OSS solutions available are: ... |
| 39 | + FIXME list alternative solution considered |
| 40 | + |
| 41 | +* Helm chart |
| 42 | +* Single read/write Master / Multi read-only slaves architectur |
| 43 | + |
| 44 | + It provides both high availability and read load balancing, without the additional complexity of multi-master configuration, which the application doesn't require at the moment. |
| 45 | + |
| 46 | +* Master failover |
| 47 | + |
| 48 | + |
| 49 | +## Infrastructure |
| 50 | + |
| 51 | + |
| 52 | +### Network |
| 53 | + |
| 54 | + |
| 55 | +#### Application Load Balancer (ALB) |
| 56 | + |
| 57 | + |
| 58 | +#### Virtual Private Cloud (VPC) |
| 59 | + |
| 60 | +* VPC subnet |
| 61 | + |
| 62 | + |
| 63 | +#### Internet Gateways (IGW) |
| 64 | + |
| 65 | +* Global egress |
| 66 | + |
| 67 | + |
| 68 | +#### Availability Zones (AZ) |
| 69 | + |
| 70 | +* 3 availability zones for high availability |
| 71 | + |
| 72 | + |
| 73 | +#### NAT Gateways (NGW) |
| 74 | + |
| 75 | +* Per AZ egress |
| 76 | +* In public subnet |
| 77 | + |
| 78 | + |
| 79 | +#### Subnets |
| 80 | + |
| 81 | + |
| 82 | +##### Public subnets |
| 83 | + |
| 84 | +* Application access |
| 85 | +* Bastion access |
| 86 | +* NAT gateways ? |
| 87 | + |
| 88 | + |
| 89 | +##### Private subnets |
| 90 | + |
| 91 | +* Database |
| 92 | +* Admin |
| 93 | + |
| 94 | + |
| 95 | +### Compute |
| 96 | + |
| 97 | +#### Nodes |
| 98 | + |
| 99 | +* EC2 compute nodes |
| 100 | + * Application nodes |
| 101 | + * Database nodes |
| 102 | + * Bastion node |
| 103 | +* Auto-scaling groups |
| 104 | + |
| 105 | + |
| 106 | +#### Orchestration: Kubernetes |
| 107 | + |
| 108 | +#### Pods |
| 109 | + |
| 110 | +##### ReplicaSets |
| 111 | + |
| 112 | +* Traefik |
| 113 | +* Application |
| 114 | +* Database |
| 115 | +* cert-manager |
| 116 | +* Metrics server |
| 117 | +* bastion |
| 118 | + |
| 119 | +##### DaemonSets |
| 120 | + |
| 121 | +* Any ? |
| 122 | + |
| 123 | + |
| 124 | +### Storage |
| 125 | + |
| 126 | +#### CSI |
| 127 | + |
| 128 | +##### EBS |
| 129 | + |
| 130 | +* For persistent storage needs (application data, logs, and monitoring metrics) AWS Elastic Block Storage (EBS) is used to provide persistent block devices to stateful pods |
| 131 | +* To integrate EBS with Kubernetes, the Container Storage Interface is leveraged by using the `aws-ebs-csi-driver` cluster addon from EKS which registers EBS volumes as a backend for K8s Persistant Volumes. |
| 132 | + |
| 133 | + |
| 134 | + |
| 135 | +## CI/CD |
| 136 | + |
| 137 | +### GitHub Actions |
| 138 | + |
| 139 | + |
| 140 | +#### Actions |
| 141 | + |
| 142 | +#### Workflows |
| 143 | + |
| 144 | +* Application |
| 145 | +* Infrastructure |
| 146 | + |
| 147 | + |
| 148 | +## Observability |
| 149 | + |
| 150 | +### Log management (ELK/EFK) |
| 151 | + |
| 152 | +### Metrics (Prometheus/Grafana) |
| 153 | + |
| 154 | +### Event and alerting |
| 155 | + |
| 156 | + |
| 157 | +## Business continuity & Disaster Recovery |
| 158 | + |
| 159 | +### TODO Recap HA features |
| 160 | + |
| 161 | +### Backup |
| 162 | + |
| 163 | +### Disaster Recovery |
0 commit comments