Simple Log Aggregation and Metrics Out-of-the-Box
December 01, 2018

Simple Log Aggregation and Metrics Out-of-the-Box

Anonymous | TrustRadius Reviewer
Score 8 out of 10
Vetted Review
Verified User

Overall Satisfaction with Amazon CloudWatch

Our engineering team uses CloudWatch to collect logs and monitor our back-end infrastructure and services. We use AWS ECS, Lambda, API Gateway, SageMaker and Step Functions; CloudWatch collects logs for these products out-of-the-box. It is easy to configure log retention policies; e.g., after three months, we can move logs to S3 infrequent-access or Glacier to save money. CloudWatch's log search in the console lacks many of the search features you would find in PaperTrail or Log.ly, but I find it is serviceable. Searching JSON-lines logs in the console might be an unpleasant experience. Similarly, CloudWatch metrics are provided out-of-the-box for all of the AWS products we use; it is easy to create alarms for these metrics and integrate them with PagerDuty.
  • Integration with other AWS products is CloudWatch's greatest feature. CloudWatch logs and metrics are provided out-of-the-box for ECS, Lambda, Sagemaker, and most other AWS products. Log aggregation and instrumentation are difficult to configure and manage; it is great to defer that work to AWS.
  • Configuring log retention policies is simple with AWS. If your business is required to retain logs for years, being able to automatically move old logs to S3 IA or Glacier with a few clicks is convenient.
  • Configuring alerts from metrics is simple, and it is easy to integrate alerts with PagerDuty or email.
  • The console's log search lacks many of the features you would find in PaperTrail or Log.ly. Regex search is either not supported, or very difficult to find.
  • It can be difficult to understand how the CloudWatch bill breaks down by log group.
  • The date/time picker in the console could be easier to use.
  • CloudWatch is integrated with other AWS products out-of-the-box; our engineers do not need to configure and maintain log aggregation services and can focus on our product. Allowing engineers to focus on building competitive advantages instead of infrastructure is a huge boon.
  • It is easy for our engineers to build alerts with CloudWatch; we know when services break and can respond before our customers discover the problem.
  • CloudWatch can be inexpensive, particularly if your business is required to retain logs for years.
CloudWatch's log search features are impoverished compared to PaperTrail's or Loggly's. However, CloudWatch aggregates logs from Lambda, ECS, API Gateway and more out-of-the-box. You do not need to manage anything. You do not need to worry about an errant logging configuration saturating your NAT's bandwidth. CloudWatch is not an APM like New Relic or Datadog, but it does provide some instrumentation. CloudWatch metrics for EC2 is most comparable to Nagios.
If you are using other AWS products, including EC2, ECS, or Lambda, using CloudWatch is an easy decision. You will get log aggregation and instrumentation out-of-the-box. The lack of log search features may be a sticking point, though your organization does not have to use CloudWatch exclusively. If your platform does not rely on AWS products, CloudWatch should not be considered.