20 أسطر
1.4 KiB
Markdown
20 أسطر
1.4 KiB
Markdown
Container Calculation
|
|
|
|
|
|
### Container Capacity Calculation
|
|
To handle 15,000 requests per second with a base container capacity of 500 requests per second and a 30% safety margin:
|
|
|
|
1. **Base Containers Needed:** 15,000 / 500 = 30 containers
|
|
2. **Safety Buffer (30%):** 30 * 0.30 = 9 containers
|
|
3. **Total Required Containers:** 30 + 9 = **39 containers**
|
|
|
|
### Cold Start Mitigation Strategy
|
|
To prevent dropped requests or high latency when spinning up new containers:
|
|
1. **Pre-warming (Provisioned Concurrency):** Maintain a buffer of extra "warm" containers running just above the current traffic demands.
|
|
2. **Health & Readiness Probes:** Ensure the load balancer only routes traffic to containers that have passed their readiness checks and are fully booted.
|
|
3. **Image Optimization:** Keep Docker images small (e.g., using Alpine Linux) so Ghaymah can pull and extract the image rapidly.
|
|
|
|
### Using Ghaymah Block Storage for Stateful Workloads
|
|
Because containers are ephemeral, stateful workloads (like databases) lose data if a container crashes. Ghaymah Block Storage solves this:
|
|
1. **Provision & Mount:** A persistent block storage volume is provisioned and mounted to a specific directory path inside the container
|
|
2. **Persistence:** The application writes data directly to this external volume. If the container is destroyed or rescheduled, the volume easily reattaches to the new container, ensuring zero data loss. |