To reduce downtime, the following strategies can be applied 1) Load Balancing: Distribute traffic evenly among servers to avoid overloading specific servers and ensure service stability. 2) Auto-Scaling: Using cloud services to automatically scale server resources in response to a rapid increase in the number of users. This makes it easier to respond to sudden demand. 3) CDN (Content Delivery Network): minimizes latency and reduces load by using globally distributed servers to deliver content closer to users. 4) Implementing a queue system: queues excess traffic in the event of a visitor spike, allowing users to access as much as they can control and preventing system crashes.

1. Logging of web server (AWS ELB access logs, NGINX/Apache2/CloudFront access logs) and associated alerts. 2. Monitoring of URL endpoints with alerts. 3. Alerts for infrastructure autoscaling events and additional alerts when maximum capacity is reached. 4. API latency monitoring with alerts. 5. Robust monitoring and other related measures.

Stronger computers Load balancing Data domain model designs Bufferize data trough medaillon architecture : allowing data consumer to access to the differents layers can help also to dispatch users In other words : having a semi coupled architecture give you more options

Pour répondre à l'augmentation de la demande, je renforcerais l'infrastructure en intégrant redondance et scalabilité, en optimisant le déploiement automatique de ressources selon le besoin. L'équilibrage de charge et un monitoring en temps réel avec alertes permettraient d'anticiper et de résoudre rapidement les incidents. Enfin, des tests de résistance réguliers assureraient des performances stables pour une expérience utilisateur fluide.

Implementing a queue system will reduce a lot of them and then probably putting some infra in place. e.g. load balancing, task prioritisation, manual intervention, work on smaller task first which will take less time and reduce more numbers, Auto scale is another thing to do, Self Service Automation to ask user few questions and narrow down