Datadog is a tool we use for monitoring of application performance, aggregating logs from our servers, and watching some metrics on each of our servers.
It's moderately expensive, so we try to balance the expense of the service with the value we get out of it.
Summary of our DataDog usage is available on this dashboard
We aggregate load balancer and application logs, but try to avoid log monitoring for services which generate vast amounts of logs - like Solr or database servers.
Pricing of Datadog is per-host, and is more expensive if that host is tracking APM (application performance.) So if you want to track things like the memory and CPU of a server it necessitates spending money. However, if you want to track metrics about one of our HTTP services you can place those checks on one of the boxes we already have and then set up monitoring in the DataDog Web API to use those metrics, which doesn't cost anything extra.
For example, to track SSL certificates configure DPUL to send those metrics -
those boxes are already sending metrics, so it doesn't cost anything extra.
There are two parts - the Ansible
configuration
and the Datadog Monitor which
tracks tls.days_left.
This is a good trick to use for many of the integrations here: https://docs.datadoghq.com/integrations/
You can get basic uptime tracking of our sites, if internal checks are sufficient, via a similar strategy above with the HTTP Check.
However, if you need cloud-based monitoring you can configure Synthetic Tests. The key to keeping it cheaper is to check as rarely as possible - for many of our sites we check from one location every 5 minutes. Figgy's manifest check configuration is an example.