t2.autoscaling - beanstalk

t2 type auto scaling by cpuCredit @ aws elastic beanstalk

functional/running test passed. but before use, Test it please

give me any report/suggestion, Welcome!

Why?

in aws EC2, 't2' type is fantastic because of the most cheapest & BURST POWER!!!
and, Elastic Beanstalk is great for manage(deploy,scaling,..) application.
I was trying to combine both, but one challenge - for auto scaling, what metric should I use?
't2' use cpu power by 'credit', and limited by that. so we can't use auto scaling by cpu usage.
aws people's suggestion was 'latency'.. but cure can't better than prevention.
So I made this - auto scaling by amount of cpu credit available

Feature

• auto scale your Elastic Beanstalk Environment (with t2 instance type only) by cpu credit
  - average of each EC2's cpu credit
• config multiple environment's scaling option
• run by lambda function (with scheduled event)
• (optional) put cloudwatch custom metric - current cpu credit(average), current EC2 scale

File Structure

• index.js - main handler & flow source. using async
• tasks.js - Detail work sources. using AWS SDK
• config.js - scaling rule configuration

How to use

1. $ git clone https://github.com/rockeee/t2.autoscaling-beanstalk.git or download zip
2. $ npm install to download npm modules
3. modify config.js for your configuration.

• 1 credit provides the performance of full cpu power for 1 minute
• Initial credit is 30(small) or 60(medium). for stable scaling, set creditThreshold_lower to lower than that.

module.exports = {
 region : 'us-west-2',
 envs :
     [
         {
          nameApp : 'My First Elastic Beanstalk Application', // application name
          nameEnv : 'myFirstElasticBeans-env',                // environment name
          creditThreshold_upper : 40,     // credit is more than this, do scale in
          creditThreshold_lower : 20,     // credit is less than this, do scale out
          scale_inc : 2,                  // scale out amount
          scale_dec : 1,                  // scale in amount
          scale_max : 5,                  // maximum scale
          scale_min : 1,                  // minimum scale
          putCloudwatch : true            // if you want to see credit/scale info, set true
         }
         // additional enviroemnt...
     ]
};

4. deploy to lamda function with your favorite method (just zip, or use tool like node-lambda)
5. check lambda function's configuration

• memory - 128MB, timeout - 10sec
• set Cloudwatch Event Rule to run your lambda function. for detail, refer this
• set & attach role to lambda function
• example role policy

 {
   "Version": "2012-10-17",
   "Statement": [
       {
           "Sid": "Stmt1453906343000",
           "Effect": "Allow",
           "Action": [
               "elasticbeanstalk:DescribeConfigurationSettings",
               "elasticbeanstalk:DescribeEnvironments",
               "elasticbeanstalk:DescribeInstancesHealth",
               "elasticbeanstalk:UpdateEnvironment"
           ],
           "Resource": [
               "*"
           ]
       },
       {
           "Sid": "Stmt1453906416000",
           "Effect": "Allow",
           "Action": [
               "cloudwatch:GetMetricStatistics",
               "cloudwatch:PutMetricData"
           ],
           "Resource": [
               "*"
           ]
       },
       {
           "Sid": "Stmt1453907067000",
           "Effect": "Allow",
           "Action": [
               "logs:CreateLogGroup",
               "logs:CreateLogStream",
               "logs:PutLogEvents"
           ],
           "Resource": [
               "*"
           ]
       }
   ]
}

6. if you config to put cloudwatch data,
   you can see Beanstalk_t2_autoscaling custom metrics @cloudwatch console.

Roadmap

• add SNS noti when scaled/failed