managing instances for the lazy
What is it?
EC2 instance creation based on user defined rules and limits.
Nothing that can't be scripted, but AWS handles some of the challenges involved for you.
Lets you lower your costs by using smaller instances during idle times, and multiplying when busy.
what i'll be touching on
basic autoscaling only - no beanstalk/cloudformation
Autoscaling Groups, Launch Configurations, and Policies
Building your AMI
Console stuff only
Where to build your group
VPC vs Classic EC2 placement
public vs private workloads
does the work actually require autoscaling?
AMI BUILDING for autoscale groups
bake-in vs configuration management
speed vs. flexibility
try to build the AMI to match the instance type you'll want
can't use network interfaces or static IPs
S3/NFS/RDS/etc may not be reachable
These are the basic building blocks for autoscaling
Located in the EC2 section of the console
full management of groups + action history
Each group lets you define :
group size (min, max, desired)
Interface is nearly the same as the process of building a regular instance.
Pick your AMI, size, security options, etc
Each launch configuration can be used by multiple groups
Scale up fast, down slow.
Keep cost in mind : charged for min 1 hour, so use the instances you have
Scaling is by a set amount or percentages :
+n, -n, =n
Cooldowns can be tricky, and will likely require some experimentation
Autoscale group metrics exist : use them to trigger your policies when it makes sense.
Notify when activities take place, at least at first.
Each group can have 0+ load balancers assigned to it.
These cannot be changed once the group is created.
Require Route53 in order to be useful.
faking my way through a q&a
and also; a tour of the console and a demonstration.
By steve bradford