1. Scaling monitoring
with
Datadog

2. Agenda
. What is Datadog?
. How does it help me scale monitoring?

3. Thank you #$%! Peter and Denise...

4. Thank you Dear Sponsor !

5. Thank you Dear Host !

6. whoami
Alexis Midon
Backend Engineer
a year in DevOps, by accident

7. environment

8. What is Datadog?

9. Datadog
. Monitoring service
. agent based
. integrated with AWS
. resource tagging

10. Datadog - cont’d
. Metrics and Alerts
. Correlation features
. Collaboration features
. Custom Dashboards

11. Event Stream

12. Default Instance Dashboard

14. nice, but...

15. How do you help me deal with:
N components: mongo, redis, nodejs, ...
x P environments: prod-1, prod-2, staging, …
x Q versions: app-blue, app-green, etc
x R users

16. help me scale!

17. How does Datadog help?

18. #1 pre-canned tools
pre-canned integrations
pre-canned dashboards

19. #2 Templated Dashboards
A dashboard can have multiple variables.
Edit once, and re-use.
$environment $zone $tier $asg ...

20. a template example

21. not bad but

22. Gimme API !!!
We can code:
. instance configuration
. infrastructure
Why not monitoring?

23. #3 Datadog has a great API.
. events, metrics, event,
tags, dashboard, alerts, …
. bindings for python, ruby, node.js, etc
. command-line

24. plain json + curl

25. using the ruby gem

27. Now, I can:
. version control my dashboards, alerts
. code my monitoring resources
. integrate with my provisioning tool

28. Integration example:
CloudFormation++

29. CloudFormation++
. a CFN template usually has related
dashboards and alerts.
. same life-cycle
e.g. app tier:
. dashboards for ELB and front-end instances
. alerts on HTTP errors, etc

30. CloudFormation++
stack = CloudFormation + Datadog
$ rake stack:app:create
executing stack:app:cloudformation:create
executing stack:app:datadog:create

31. in git
/stacks
/app
app_cfn_template.json
app_datadog.rb
app_http_alerts.json

32. be creative
datadog.rb is evaluated in a rich context.
It has access to everything.
very flexible.

33. Cons / Pain points :-(
. still have to deal with some json
. room for drift - if users manually edit
resources
. resource tracking can be tricky

34. Pros
. monitoring has code
. all the benefits of using code:
tests, versioning, tracking, DRY, bugs, ...

35. Summary
Go code your Monitoring,
with the awesome Datadog API.

36. impressed ?

37. Thank you!

38. Scaling monitoring
with
Datadog