How many uploads can my server handle ?

We wouldn’t know as it’s application/platform dependent. You’d need to run a benchmark and find out, ApacheBench (ab) can help you with this.

Heya,

The number of simultaneous uploads your server can handle depends on several factors, such as:

1. Network Bandwidth: If many users are uploading simultaneously, it could saturate your droplet’s available bandwidth, especially if each user is uploading multiple megabytes of data.

2. CPU and Memory Usage: Handling uploads, especially decoding and processing base64 strings, can be CPU-intensive. Additionally, large uploads could increase memory usage, especially if your application is buffering the entire image in memory.

3. Disk I/O: If you’re saving these images to disk, the disk’s I/O capacity can also become a bottleneck.

4. Gunicorn Workers: With 4 workers, theoretically, you can handle 4 simultaneous requests. However, this doesn’t mean you can only have 4 users, because the process of uploading involves a period where the server is simply receiving data and the worker is essentially waiting. Depending on the real-world scenario, a single worker might be able to handle multiple uploads from different users if these uploads are staggered.

To answer your question:

• For 10 users: It’s likely that your server will handle this without any issues, especially if the uploads are not all occurring at the exact same second.

• For 50 users: It starts to depend on how simultaneous these uploads are and the aforementioned factors. If all 50 users try to upload at the exact same moment, you might see some slowdowns or even timeouts. However, if their uploads are staggered even by a few seconds, it might be manageable.

• For 100 users: This would definitely push the limits of a $10/month droplet, especially if uploads are near-simultaneous. You may see noticeable slowdowns, timeouts, or even out-of-memory issues.

Recommendations:

1. Upgrade Your Droplet: If you anticipate a sharp increase in users, consider upgrading your droplet temporarily to handle the load. You can scale down later if needed.

2. Monitoring: Make use of monitoring tools like htop, iotop, and iftop to monitor CPU, disk I/O, and network bandwidth respectively. This will give you insights into potential bottlenecks.

3. Load Testing: Consider using tools like locust or ab (Apache Benchmark) to simulate multiple users uploading data to your server. This will give you a better understanding of how your server reacts under load.

4. Optimize Flask & Gunicorn: If not already done, ensure that Flask is running in ‘production’ mode, and consider increasing the number of Gunicorn workers if CPU and memory resources allow.

5. Content Delivery Network (CDN): If your project grows further, consider using a CDN or a service like Amazon S3 to handle uploads, which can alleviate the load on your server.

6. Error Handling in Your App: Ensure your web app gracefully handles errors. If the server becomes overloaded and starts returning 5xx errors, the app should ideally retry the upload after waiting for a bit.

Finally, thank you for working on a humanitarian relief project. Such initiatives make a big difference, and I hope your technical setup supports your mission effectively!