Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I'm not exactly a system administrator by profession, so this is a foggy area to me.

From what I know, AWS assigns every new PC a random ID, which is quite hard to work around in most software. So what I'm thinking is, is to create a Linux cluster with 3/4 PCs, one with static external IP which also works a DHCP and DNS server, so that all new PCs just get adresses out of it's static allocated pool.

Is this possible with AWS?

share|improve this question
Almost certainly not. Can you elaborate on what you're trying to do, when you say "quite hard to work around in most software"? – SmallClanger Jun 11 '11 at 13:28
I need to set up cross node communication based on DNS names, and they need to be set up in config files, which creates ton of problems once one of the nodes restarts and gets new random an cryptic DNS name. – Coder Jun 11 '11 at 13:34
up vote 4 down vote accepted

Within the framework of AWS, I think you've got three options:

  • Setup a Virtual Private Cloud. Once you've got your own internal network, you may well be able to run our own DHCP and manage the network accordingly. I can't say what this will do to the cost, though.

  • Give each instance an elastic IP and its own DNS, having them communicate via those. This might incure extra bandwidth costs, though.

  • Set up another, EBS-backed micro instance to act as a messaging server between your worker instances.

The last one feels to me like an option that most fits the AWS model. Your worker agents can be entirely stateless and rebooted on a whim, where they'll check-in to the (persistent) messaging server to announce their presence and be ready to start work. The other two will wokr, bu you're essentially running three of four persistent, EBS-backed vitual machines, and AWS isn't the cheapest way of doing that, especially if you have a high compute/bandwidth requirements.

share|improve this answer
Instead of having to manage your own "messaging server", why not just use SNS/SQS? That's what we use for our workers then we don't have to worry about the messaging node. – Smudge Jun 11 '11 at 15:03

Yes it is.. You have to use their VPC offering. THen you can choose to add a subnet and when you create EC2 instances you can give them an IP on that subnet. So some nodes might not even have a public IP.

From there you can do a site to site vpn connection from amazon to your office/other datacenter to make the two lans talk.

share|improve this answer

I think Amazon Virtual Private Cloud (VPC) is what you need. Alternatively, you can just use security zones in order to isolate your virtual servers into separate groups and restrict traffic between them.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.