0

I have a WordPress site currently hosted on an EC2 server. After doing some digging on the interwebs, I found Spot Instances - often half the cost of a normal instance.

The problem I see with it is that the Spot instance could terminate at any time. I think I've come up with the solution (some downtime is OK) but not the implementation.

You see, my idea was to have a spot request all the time - at a certain price. When the request is fulfilled, my on-demand server would be backed up (S3?) the spot instance restored to that backup, static IP's switched, and the on-demand killed. When the Spot is running, it could back up its contents every so often (site mostly static). When the Spot is terminated, a new on demand would be started, restored to the Spot's backup, get the IP and continue life.

My problem lies in implementing the process. I don't know of any tools that would allow this kind of monitoring and action.

Any help?

3
  • 1
    While I agree with the comments below, that the reserved instances might be best for your situation. If not, this should work... 1. Create you server, build your website, create an AMI. 2. Power off that server, spin up a new spot instance using that AMI. 3. Create new AMIs every N number of days via the API. 4. Setup CloudWatch to monitor that server, if it goes "away", spin up a new spot instance with a higher bid price.
    – af-at-work
    Mar 27, 2013 at 20:21
  • @af-at-work Make that an answer and I'll accept and up vote.
    – user162327
    Mar 28, 2013 at 17:14
  • @xxx What are you looking for in an answer that isn't already covered by the existing two?
    – Ladadadada
    Jul 16, 2014 at 14:10

3 Answers 3

3

Are you aware of Reserved Instances?

They are also significantly cheaper than On Demand instances but have the added benefit of not disappearing halfway through processing a request.

1
  • Truth. A Heavy Usage Reserved instance is a very significant discount versus the standard hourly rate.
    – sysadmin1138
    Mar 26, 2013 at 22:40
1

While I agree with the comments below, that the reserved instances might be best for your situation. If not, this should work:

  1. Create your server, build your website, create an AMI.
  2. Power off that server, spin up a new spot instance using that AMI.
  3. Create new AMIs every N number of days via the API.
  4. Setup CloudWatch to monitor that server, if it goes "away", spin up a new spot instance with a higher bid price
0

EC2's primary advantage is the ease of starting up new instances rapidly, and of automating that. It's great for:

  • large services where you fire up more or less servers according to demand.
  • very rapid deployment where you have an emergency need for a new server.
  • Short term requirements like firing up a temporary batch of systems for doing some number crunching.
  • Storage pricing might be quite competitive if you have a lot of it.

However, it comes at a cost premium. If you just want to host a blog, you'd be better off with a more traditional hosting solution, which will come at a fraction of the price. (Eg I use a Virtual Dedicated Server from hostsense.net, for $10/month, plus a backup service) I expect moving your operation to such a service will also be less work than setting up automated systems to move your server between Amazon instances as you suggest.

You must log in to answer this question.