|
|
|
|
|
by mschuster91
1543 days ago
|
|
> Honestly, most of these companies would be better off using Google, Azure or AWS' SSO-as-a-Service product (if that's what you're hoping to get out of Keycloak). The thing is, your Keycloak instance is not going to matter to any hacker, particularly if it's inside a VPN and not reachable from the Internet - and while we're at it, fuck zero-trust because it is essentially the same level of stupidity as using Okta, you're once again putting all your eggs into the basket of whatever provider you choose. Your SSO-as-a-service provider however? They're the juiciest target out there that is. Everyone from secret services over enemy nation states to your average cyber-criminal is looking to get access there. And as we've seen, all it takes is a couple teenagers and a couple thousand dollars. Good network design costs a lot of money to set up, particularly to limit the scope of an attack (e.g. because the VPN software had a vulnerability), but it's orders of magnitude better in the long run than to outsource core IT to some incompetent fools with subcontractors. |
|
This doesn't make it particularly usable as SSO...
>Good network design costs a lot of money to set up, particularly to limit the scope of an attack (e.g. because the VPN software had a vulnerability), but it's orders of magnitude better in the long run than to outsource core IT to some incompetent fools with subcontractors.
This is exactly my point. Most businesses not not have the resources to maintain this level of infrastructure.
Additionally, I'm personally of the opinion that walled gardens with VPN entry points are a particularly good choice for modern businesses these days. Even the White House OMB is pushing the beyondcorp model in their recent recommendations for ZT.