Anybody working regularly with IaaS providers such as Amazon or Rackspace can recount a personal story of a forgotten instance.

The most dramatic stories are not about a cheap micro instance – my personal story with AWS cost me some US$200 when I missed to turn off an EC2 instance and went for a diving trip to Egypt. I’ve got a number of suggestions that might save you some money.

• Above all, you want to avoid paying for unused resources. Using auto scaling is a great mechanism for running only the required instances, and for example, to scale down at night when fewer EC2 instances are required.
• Often the monthly bill tells you that there something is still running somewhere. Make sure you stop unused resources as quickly as possible. If you know in advance that you want them to be stopped at the end of the day, then use the Unix at command to schedule the termination of the instances.
• Although AWS management console provides dashboards, there is no super-dashboard. Instead, you have to flip through all tabs yourself (starting from from the “S3” tab to “EC2” and all the tabs up until “RDS”). Only after checking all tabs can you be sure you have an accurate overview of the current resources for the selected region.
• Remind yourself that the AWS management console is always displaying resources per region. Once you switch to another region, e.g. from Asia/Pacific to Europe, you will no longer be able to see instances running in Asia/ Pacific.
• The console is sometimes out of sync. When this happens, remember to click on the refresh button so as to avoid only seeing outdated information.
• The command-line tools I introduce later will work with resources for the default region in the US (unless you specify otherwise). Remain vigilant at all times e.g. when working in Europe do not start and then forget an instance in the US.
• Always double check for running instances before engaging in another project, leaving for a sabbatical or a trip around the world.

Be careful and make sure you don’t wast money that is better spend for a fabulous diving trip.

These tips are taken out of my Middleware and Cloud Computing book.

Update as of May 2012:

You can setup billing alerts now for AWS and use SNS to recalculate your auto scaling. See Jeff’s posting on AWS typepad here.

Quite often I chat with customers about cloud computing and tell them they could have a server running in the cloud for as little as 1 cent per hour.  Then their next question is: “Well, if it is running in the cloud, how can I access it then?“.

There seems to be a lot of confusion about it. Just because your server is running in the cloud, doesn’t mean it is not tangible. So the short answer is: “Use ssh and connect to it.“. Your server is not hidden in a nebula – it’s fully connected to the internet (at a fantastic bandwidth, a WebLogic 10.3.2 package download is completed in 4 mins at 3.8 MB/s) . For Amazon’s AWS you typically have to provide a X.509 certificate and RackspaceCloud uses password for root login with ssh.

Windows Options

Even if your managing the cloud from your Windows desktop you are not out of luck connecting to a UNIX server in the cloud, there are a couple of options:

• There a free version of VNC for private and commercial use which is called TightVNC, see http://www.tightvnc.com.

• NX is a desktop virtualization solution from a company called NoMachine. NoMachine designed the NX libraries to provide a self-tuning protocol with compression and reduced round-trips for X11 even over slow connections. Have a look at their homepage under http://www.nomachine.com. They provide a free version which is limited to two connections only.
• FreeNX was created in 2004 as the first complete free and GPLed server implementation of NX, see http://freenx.berlios.de.

• Cygwin is another solution, see http://www.cygwin.com. It provides a UNIX environment for Windows and comes with a X-server that can display your X based applications on your UNIX host. The software was written by Cygnus Solutions which now belongs to Red Hat.