-
Notifications
You must be signed in to change notification settings - Fork 54
AMI launch
The purpose of this guide is take users with no prior experience in Amazon's AWS/EC2 cloud infrastructure from zero to running experiments with NEXT on an EC2 machine.
There are five steps:
- Setting up your AWS account
- Set up an AWS access key
- Set up an AWS key pair
- Starting a computer on EC2 running NEXT
- Starting an experiment on NEXT
For the first three steps, Amazon also provide their own getting started guide, so users seeking further detail on EC2-specific topics should look there.
- Create (or otherwise possess):
- An AWS account
- An AWS access key ID, AWS secret key (needed after NEXT starts)
- A named AWS keypair
- Start an EC2 machine with the following specifications:
- Amazon machine image (AMI) with the ID specified here
- Machine type
c3.8xlarge
or better - At least 100 GB of storage
- Ports 80 and 8000 open to the world (or to only the IPs of every computer you want to allow, if you know these in advance and care about this)
-
Go to
http://aws.amazon.com/
and click Sign Up. -
Follow the on-screen instructions until you are logged into the AWS console.
-
Go to the Security Credentials in the dropdown under your name.
-
Click on the Access Keys (Access Key ID and Secret Access Key) tab.
-
Click the create new access key button. In the corresponding pop-up, click show access keys. Write your Access Key ID and Secret Access Key down or save them to a secure location for later reference.
-
Go to the AWS EC2 console. If you are somewhere in the console, return to the main page by clicking the orange cube at the top left and then click "EC2" on the main menu page (should be around the top left).
-
Click on the Launch Instance button
-
Select Oregon as the region in which to launch (otherwise the AMI can't be found)
-
Select Community AMIs on the left-side panel and use the search box to search for the AMI ID show below. Once you find it, click Select on the right.
- Be sure to find the AMI with the ID ami-36a00c56. This AMI pulls the latest release of NEXT.
-
Select a machine type to suit your needs. Active learning experiments require a lot of computation resources. We recommend at least type
c4.8xlarge
which has 32 cores and 60GB of memory if you plan on running an experiment that has a large number of participants. Then click Next: Configure Instance Details. -
If money is no object for you, click Next: Add storage. Otherwise, there is a potentially cheaper option accessible from this screen, but with several caveats. To read about this and its potential ramifications, click here. Click on Next: Add Storage
-
Expand the root volume to at least 100 GB by locating the box under "Size (GiB)" and placing in there a value at least 100. Then click Next: Tag instance
-
Give your instance a name (what name you choose isn't particularly important, but make it different from any other instances you have created). Then click Next: Configure Security Group
-
Click Add Rule twice. In the Port Range boxes for the two new rules, enter
80
in the first, and8000
in the second. Make sure the Source says Anywhere (unless you know what you are doing, in which case populate the sources with the subnet you wish to grant access). Then click Review and Launch. Of you left your source as "Anywhere" a warning may appear about your permissions. This is expected and not problematic. -
Click Launch. In the dialog that appears, select Create a new key pair. Name this key pair and download it. Do not lose or share the downloaded file! Once you have downloaded it, click Launch Instances.
-
Click View Instances button.
-
Locate the instance by name in the list of instances that comes up and click it. This will bring up the instance details in the lower panel. In particular, the text after Public DNS is the URL of the machine. Copy this and paste it into a new browser tab to see your NEXT dashboard. This may take up to 20 minutes to come online after the instance indicates a status of "Running".
- If your dashboard is not coming up and you have waited a good 20 minutes, right-click your instance in the instance list on the AWS console and select Reboot