1and1 Cloud hosting scaling of DB and Instances

Question

do have anyone experience with 1and1 hosting? I've tried to read over their "transparent pricing".

They have a configuration panel, that allows to select amount of CPU/RAM/SSD storage.

(Configuration Panel)

What is that configuration about? And how do autoscaling work?

• The configuration is "per-machine" and if the machine is too busy they instantiate another machine?

• The configuration is "total hardware power" divided among machines? (in that case then the only thing that scales is the price since traffic load may vary according to time of day)

• If I choose the most powerfull hardware combination, and I have just 1 user at any time, will I pay the full computing power? (In example the configuration is able to host 1.000.000 users, but for some reason the server is kept busy 24h/24h just by 1 user)

• Do I have to configure a different DB for each "machine"? If yes how do I assign databases Usernames/Passwords and IP adresses.

• Do computing power scale automatically? If yes why they mention I will be able to change configuration in 55 seconds at any time? Why do I need to change configuration if the Cloud "scales"?

I know that are lot of questions, but in reality it is just one question: How do 1and1 Cloud hosting scales?

Answer 1

The configuration you select is per machine. So when you specify the spec's for what you are after it is the spec's per machine being spun up. The way 1and1 scale up is two fold. First of all you have what is known as burst capacity. With burst capacity there is the ability to go over the spec's you have specified by a small amount (can't find the exact burst spec's at the moment) for very short periods of time. This shouldn't be considered a means to bypass pricing but rather a way to stay online should something happen where you need added capacity for a second or two. In the other way you can also have multiple machines spun up to the same specification and balance the connections across all of them.

If you provision a single server with a very high hardware profile then that will meet your needs but will be kept active even by a single connection. It is generally better to go for a lower capacity machine and scale horizontally (that is balance the connections across a few machines) while you are getting a feel for what sorts of capacity you need

Whether you need to configure a different db for each machine depends on what you are using the machines for. If you are wanting a single db and each machine will be connecting to the same db then there is no need to configure a separate db for each machine, however if each machine will need its own db for some reason (such as a shared hosting environment as an example) then yes each machine will needs its own db. Once again it depends on what you are using it for.

As for scaling computing power the concept of scaling here is that you can increase computing power within a relatively small amount of time without needing to terminate the instance and provision a new machine with higher spec's. The reason why they say it takes 55 seconds is because their back end systems that perform the increasing of compute resources take that long to provision the additional resources and assign them.