I would classify things like NFDB, or any large infrastructure as application specific.
An AS system is usually not useful to anyone except the people it developed it, and the problems it solves at that place. Those usually require hitting a certain scale (one that is low or high depending on the type)
Anyone else that has a similar problem will not be well served by someone else’s AS system, as “similar” is not enough. And they usually have the skills to build their own.
The only people that benefit from open sourcing such a system is people that want to rip off existing effort for a quick buck and don’t care how well it runs or fits them.
For the average user, there are the standard options, that work much better at their scale.
Login to reply
Replies (1)
> Infrastructure, and usually code powering them is very hard to open source imo. In a way that is useful.
> or any large infrastructure as application specific.
Your servers and your network is not my servers nor my network. Your control plane and service discovery (if you have one) might be meant for colo, full cloud, kubs as a service, physical hardware, theres' no point releasing it.
I think many don't admit there is some security through obscurity as well. Knowing long-term topology could allow an attacker to cripple systems more easily
My block storage might be cheaper than yours, my object storage might be cheaper. There are so many differences.
For example,
- my new infra relies on a low latency control plane now, that can't really happen in the cloud
- I have shared storage on a trusted network
just heavily depends on the application and costs involved.