CloudWatch-ing:

Why should you listen to me?

Over 1 Year Playing with CloudWatch

Visibility was my 1st real project

I could gush about this for DAYS!

Story Time

Back in the day...

Things were good...

...or so we thought.

Great Debacle of 2018

We were definitely doing it all wrong

Seriously.

Enter AWS CloudWatch!

What & Why

Where & How

Before CloudWatch

After CloudWatch

My Favorite Things

Ease of Use & Integrations

Centralized Logging

Searchable Logs

Cross-App Metrics

SHOW & TELL

Terraform Metrics

Basic Metric Filter

							resource "aws_cloudwatch_log_metric_filter" "unique-name-of-your-filter-resource" {
							     name    = "name-of-your-filter-pattern"
							     pattern = "\"the actual pattern you are looking for\""
							     log_group_name = "logroup/app.name.orwhatevs/"

							     metric_transformation {
							          namespace = "${var.metric_namespace}"
							          name  = "Easy-To-Spot-Name-Of-Your-Metric"
							          value = "1"
							          default_value = "0"
							      }
							}
						

Parsed Metric Filter

							resource "aws_cloudwatch_log_metric_filter" "unique-name-of-another-filter-resource" {
							     name    = "your-new-filter-pattern-name"
							     pattern = "[..., user=*root*, spacer=*-*, colx=*Some* && coly=*Web*, count]"
							     log_group_name = "logroup/app.name.orwhatevs/"

							     metric_transformation {
							          namespace = "${var.metric_namespace}"
							          name  = "Easy-To-Spot-Name-Of-Your-Metric"
							          value = "$count"
							          default_value = "0"
							      }
							}
						

Basic Metric Alarm

							resource "aws_cloudwatch_metric_alarm" "unique-name-of-your-alarm-resource" {
							     alarm_name = "Easy-To-Spot-Name-Of-Your-Metric"
							     alarm_description   = "10 errors in 10 minutes Error connecting to SQL Server"
							     comparison_operator = "GreaterThanOrEqualToThreshold"
							     metric_name         = "Easy-To-Spot-Name-Of-Your-Metric"
							     namespace           = "${var.metric_namespace}"
							     period              = "900"
							     statistic           = "Sum"
							     threshold           = "10"
							     evaluation_periods  = "1"
							     datapoints_to_alarm = "1"
							     treat_missing_data  = "notBreaching"
							     alarm_actions       = [
							          "${data.aws_sns_topic.yada.arn}","${data.aws_sns_topic.blah.arn}"]
							}
						

Resources

CloudWatch Documentation

Terraform with CloudWatch

Grafana Documentation

Major Keys

Major Keys

Log Levels Matter

Centralize Your Logs

Log Strategically

Send Alerts to People

Metrics Make a World of Difference

This could be YOU!