The freemium service provides access to compute-heavy machine learning models running on GPUs.
Hackers blast 50-100 requests in the same second, which clog the servers and block legitimate users.
We reported IPs to AWS and use Cloudflare "Super Bot Fight Mode" to thwart attacks, but the hackers still break through.
We don't require accounts, but could impose account requirements if this helps.
Any suggestions?
The one event that cannot be automated is cursor movement/position. Put a check into your event handlers that check that the cursor is actually over the event target.
Are you saying block form submission unless the cursor is over the event target?
If so:
* How to handle legitimate requests from mobile users?
* How to handle form submissions with the "return" key?
Return key on a control in a form will fire a submit event. Check for cursor position in your submit handler.
Are there proxy examples you could point us to?
Thanks for your help.
The freemium service provides access to machine learning models on GPU instances, served with FastAPI.
Each request invokes a compute-intensive ML model, but perhaps there is something wrong with the FastAPI configuration as well?
I watch the FastAPI repos a lot and tones of people do not understand how async python works and put their models with sync code in an async context.
We tried removing "async" -- thinking it would force sequential processing -- but it unexpectedly seemed to cause parallel processing of requests, which caused CUDA memory errors.
Before removing "async", this is the weird behavior we observed:
* Hacker blasts 50-100 requests.
* Our ML model processes each request in normal time and sequentially.
* But instead of returning individual responses immediately, the server holds onto all responses -- sending responses only when the last request finishes (or a bunch of requests finish).
* Normally, request 1 should return in N seconds, request 2 in 2N seconds, but with this, all requests returned in about N50 seconds (assuming batch size of 50).
1. Any suggestions on this?
2. Mind clarifying how sync vs aync works? The FastAPI docs are unclear.
Any help would be much appreciated.
This has been extremely frustrating.
The memory errors you're seeing could suggest that you may not actually be able to run multiple instances of the model, and even if you could it may not actually give you more performance than processing sequentially.
Seems like ultimately your current design can't gracefully handle too many concurrent requests, legitimate or malicious - this is a problem I recommend you address regardless of whether you manage to ban the malicious users.
@headlessvictim2 search for "Asynchronous Request-Reply pattern" if you want more information about this kind of architecture. You will remove any bottleneck from the API server and can easily scale out from the task queue.
How would this work with GPU-bound machine learning models?
The model processing takes > 30 seconds and would still represent the bottleneck?
Here is an example of what it could look like: https://docs.microsoft.com/en-us/azure/architecture/patterns...
Right now, we use ELB (Elastic Load Balancer) to sit in front of multiple GPU instances.
Is this sufficient or do you suggest adding Celery into this architecture?
There is an event loop that goes through all the tasks and runs them.
The issue is the event loop can only move on to the next task when you reach an await. So if you run a lot of code (say an ML model) between awaits no other task can advance during this time.
This is why it is co-operative, it is up to a task to release the event loop, by hitting an await, so other tasks can get work done.
This is fine when you have async libs that often hit awaits at things that are IO related like say db, or http calls.
FastAPI will spawn controllers that are not defined as async functions on a thread pool but it is still a python so GIL and all that.
You should do as the sibling comment says and decouple your http from your ML and feed the ML with something like Celery. This way your server is always there to respond to things (even if just a 429) to hit a cache or whatever else.
Right now, we use ELB (Elastic Load Balancer) to sit in front of multiple GPU instances.
Is this sufficient or do you suggest adding Celery to this architecture?
It appears less like malicious DDoS and more like pragmatic theft.
It is possible, but this degrades the experience for legitimate users.
We prefer solving this without impacting/taxing normal users if possible.
edit: Actually, since you use CF, just make a firewall rule that forces the captcha for those ASNs before it even gets to your app. They have a field named "ip.geoip.asnum" for that, and an action called "challenge" which will force a captcha.
It is possible, but this degrades the experience for legitimate users.
We prefer solving this without impacting/taxing normal users if possible.