0

I need to run a custom-made software authentication system in the cloud and I was considering buying an Amazon EC2 (probably t2.micro) instance on demand. I will also need an elastic IP address since users will have to find my instance through a DNS.

What I'm wondering right now is the following: I couldn't get how it works with the on-demand model. Will the machine be put to sleep when there are no requests as it happens with the red-hat free cloud? Will that then require several minutes to restart? (I can't afford users to wait for their software to be authenticated for several minutes)

And finally: will the elastic IP address wake up the machine when a request comes?

I could find no piece of documentation regarding these issues

Improve this question

2 Answers 2

Reset to default
1

An on-Demand instance stays on as long as you don't shut it down or reboot it.

The only difference with a physical server (hardware) you have on premises is that you don't need to buy it. And there is no commitment to use it for 1 year, 2 years, 3 years or even a day. That's where the On-Demand term comes from. You can turn it on or down as you like, when you want it or need it.

Improve this answer
1
  • To further clarify: "on demand" is a billing term, as is "spot" and "reserved instance". Spot instances are very cheap, they're AWS trying to get some money for spare capacity, but can be shut down any time AWS wants with virtually no notice. "On demand" and "reserved instances" will not be shut down by AWS under normal circumstances, only for things like hardware failure, which of course is rare.
    – Tim
    Mar 6, 2017 at 19:48
1

On-Demand instance is on (and you are billed for those hours) from the moment of start until it is stopped or terminated. It is not in 'sleep' mode even if there's no workload. It is really just billing term.

Regular external IP will be changed if you restart the instance (but not if you reboot it). If you need to keep static public IP then use elastic IP: one such IP per instance is free of charge as long as the instance is running.

Why use on-demand? You don't have to pay upfront anything. If after a week from your start you found t2.micro is too c,mall (or too large) you can scale the instance anytime and you'll be billed according to the new instance size.

Pls note you're always billed for full hour even if you used two minutes from it.

For burst webservers: t2 type was made for this purpose but don't forget to watch CPU Credit Balance at monitoring otherwise your performance may be unexpectedly low.

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .