Choosing a cloud service for a specific task

Question

Cloud computing

I ask for help in finding a suitable solution from the pros.
Task.
1. We need several Windows applications constantly running with an interface that need access to the Internet. These applications do not require incoming connections. It is highly undesirable to restart applications and or the virtual machine as a whole.

2. I would like to have access to the server via RDP with the preliminary inclusion of this access.

3. I would like to have access to the server through some kind of console.

4. If the server has rebooted, then I need to find out about it, for example, by letter. Better - by letter and SMS. I don’t know what to do in case of a shutdown, but I would also like to know about it, if this is even possible. In addition, I need all my applications to start automatically after reboot in the order I need with a given delay. I suppose that launching applications after a reboot can be implemented with a script.

5. It would suit me if every time I need access to the server, I would go to the Windows Azure portal, include a real IP there, then create an endpoint and then get access to the server via RDP. That is, I do not need a real IP all the time. Maybe I just don't know how to get rid of it.

6. Minimum payment per month.

What and how of the desired is implemented now.
1. Windows Azure. One virtual machine. It runs the required applications. There is no reboot control. I don't know how best to implement it.
2. Disadvantages:
2.1. The server is visible directly to the Internet, because of this, constant brute force from the Internet. I also pay for this traffic, if I'm not mistaken. certainly. I just (while writing) came to the conclusion that you can delete the endpoint on the Windows Azure portal and add it only when you need access via RDP, and after the RDP session, delete the endpoint. In this case, brute force will only occur during periods when the endpoint exists ... but the server will remain "visible" on the Internet anyway.
2.2. No console access. If something is wrong with the server (for example, there is no RDP access for some reason), then I will have to download the VHD file and work with it locally. I don’t know if other companies have other solutions, but such a solution is not very convenient, IMHO.
2.3. There is no possibility to refuse the real address at all. Or I don't know how to do it?
3. Benefits: I pay $12/month on a As-You-Go plan, which is quite reasonable for me.

Please help me to achieve the desired result. Perhaps I don’t understand something, and besides what I wrote, it is necessary / possible to implement something else.

I would be very grateful for any help. I can offer in exchange for comprehensive assistance - material rewards in a peculiar form. More precisely, I will explain to someone who will undertake to help.

Answer 1

I will comment on Azure:
1. The availability of the machine is provided by a regular HA cluster, which means a hard reboot when the server fails.
2. Through the creation / removal of "endpoints", preferably on a non-standard port. Or restrict the IP addresses from which RDP is available through the built-in firewall.
3. There is no console, it is proposed to use backups to recover from failures .
4. There really is no monitoring of reboots, notifications have recently appeared , but the metric you need has not yet been added. You can implement the function in one way - by setting up monitoring inside the VM.
5. You do n't have it , the server is accessible from the outside only through "endpoints".

Answer 2

and Azure, you can set up two virtual machines in the local network, one of these virtual machines (Linux) can be exposed to the second interface on the Internet.
On Linux, open access to anyone from where you need + vpn, if desired. Again, a minimal Linux instance is unlikely to eat much.
PS By the way, Azure considers that you have at least two hosts doing the same task, so you should keep in mind that there may be downtime.