When it comes to the protection, I tested combahton when I was with deinserverhost.de and the protection was "okay"I have used KMS host back then quite extensibly for teamspeak and other stuff. I was very happy with the service. There was only one mishap by the support once but that was compensated for properly. Other than that the uptime was absolutely solid with only few scheduled maintenances. As for the anti ddos solution, there are better people to ask than me.
Hi Joseph from fastpipe.io here - glad to clarify what we really use for our SSD Cloud Servers
Our SSD Nodes consist of HP WS460c Gen8 / BL460c Gen8 and BL460c Gen9. CPUs are E5v2, E5v3, E5v4 - SSDs are mostly from Sandisk, Samsung or Crucial (two different vendors per Node).
Ram is overall 12x 16GB or 16x 16GB DDR3/DDR4 ECC - we do buy them in packages of nearly hundred sticks per delivery. Connectivity is done with HP Flex-10 modules by 2x 10Gbit per physical node, each bladecenter has at least 2x 10Gbit to our routers.
The following is a recent photo from one of our bladecenters in Frankfurt, at our FFM2 site:
As you can see, two ssd drives per server, no old server hardware from 2009 or so. It wouldnt even make sense to operate such old hardware, as we would pay nearly the triple for power usage as for the hardware shown on the picture. We do around 3,6kW power draw per full loaded bladecenter. I remember the time quite well, when we did the same with eight servers. Nowadays it's not longer reasonable to operate servers which draw 400-500W load for a fraction of the processing power of newer ones.
Our physical nodes are monitored closely via both Icinga2 and Grafana. We dont do overselling and let at least ~15-25% of spare ressources available on every node. To do so, our automated deployment choses a node depending on how much cpu, memory and storage resources it has. This happens once you have added money to your customer account and your server is provisioned. The below image represents the recent load from one of our nodes (20 E5 Cores - Dual CPU, 192GB Ram, 2x 2TB SSD - Hardware Raid 1 with 1GB Cache + BBU):
View attachment 2072
Nearly 80% overall load, no IOWait and much available sparse ressources
If you have experienced a bad performance, it can be related to a running backup which is done overnight. We do use snapshot based backups, which do copy-on-write. Means your server does not need to be shutdown and it's data is "replicated" on our backup infrastructure in real-time. You can simply turn off backups in our customer area and do them on your own.
try fastpipe and i am sure nothing can be bypassedWhen it comes to the protection, I tested combahton when I was with deinserverhost.de and the protection was "okay"
When you are experiencing a proper attack, the protection is useless. Up to 3 gbp/s it was quite okay, although the latency increased by 200ms
The most unique DDoS protection I have ever seen was on vpsrv.com and their servers. They use OVH but also do their own software firewall and configuration, which even held back an attack of 98 gbp/s without the latency increasing by more than 10 ms
before buy a dedicated server make sure you give a test on a small cloud server and then buy itIf I'll be in search for new provider, maybe I'll buy dedicated from them to give them a try