OVH is known for having good DDoS mitigation, ridiculously understaffed support and great pricing. I think all three of those things make it appealing for that niche.
OVH DDoS protection makes it quite popular for hosting game servers too thanks to kids paying for DDoS attacks.
if one wanted to host a critical app with beefy db requirements on baremetal, proxying through cloudflare for ddos mitigation, preferably in north america, do you have any suggestions on where to look? i used to think "http://packet.net" but not sure abt "equinix metal"...
We OVH with their Canadian location (Beauharnois) for nearly all the GrapheneOS servers and it's my overall first choice for anything. I've used them for many years and have always found it reliable. The stories about their support being terrible are definitely true though.
DDoS mitigation from their network and unlimited bandwidth are nice. I wouldn't use their US location since it's much smaller.
They have terrible user interfaces, terrible support and some of their newer service offerings like their domain registrar / DNS are a bit primitive.
I don't want anything to do with AWS or OpenShift again. I'm quite happy using OVH for everything other than DNS. I'd like to use it for DNS to have everything in one place once they improve it. Don't use Cloudflare anymore because my experience is it's slow and not very helpful.
You really don't want to put non-website HTTPS services behind Cloudflare because they end up silently breaking them with the challenge pages.
It will break web resource downloads too if they aren't being made alongside the main document requests.
It's also just really slow...
https://cloudflare.com/en-gb/products/cloudflare-spectrum/… probably works a lot better for anything that's not a website even if it's HTTPS.
Their reverse proxy for web sites is just weird. It does some really sketchy things partly because nginx is somewhat sketchy and they've hacked it all together on top of it.
wow thank you !! this was all very helpful !! w.r.t. ovh, is their support lacking for datacenter-wide or network-wide issues, or does it pertain more to individual hardware issues?