[hintymad]

AWS frontend services are usually implemented in Java. If Kinesis' frontend does too, then it's surprising that the threads created by a frontend service would exceed the OS limit. This tells three possibilities: 1. Kinesis did not impose a max thread count in their app, which is a gross omission; 2. Or there was a resource leak in their code. 3. Each of their frontend instances stored all the placement information of backend servers, which means their frontend was not scalable by backend size.

[joneholland]

My understanding is that every front end server has at least one connection (on a dedicated thread) to every other front end server.

Assuming they have say, 5000 front end instances, thats 5000 file descriptors being used just for this, before you are even talking about whatever threads the application needs.

It’s not surprising that they bumped into ulimits, though as part of OS provisioning, you typically have those tuned for workload.

More concerning is the 5000 x 5000 amount of open tcp sessions across their network to support this architecture. This has to be a lot of fun on any stateful firewall it might cross.

[foota]

The tcp connections are probably not an issue, working in cloud it's never something I've seen worried about, so maybe the architecture doesn't have that limitation?

[joneholland]

Any time you traverse a firewall or a NAT you’ll run into issues if you have a massive amount of open sessions and you are dealing with a stateful FW.

Just because you haven’t encountered it doesn’t mean it’s not there, it’s probably just properly tuned and balanced for the load.

[foota]

Yeah, I believe that. I'm not sure this applies to something like AWS though, where firewall like capability is provided via a layer spread over thousands or greater of instances.

[joneholland]

The virtual firewalls running on the virtual networks in AWS for their customers are not the same as the layer 2 firewalls that exist in their data centers internally.