We have been considering how to best handle internal monitoring (i.e. internal stats like disk space, CPU utilization). Here's our current set of solutions:
References:
- https://www.pagerduty.com/integrations/ (Infrastructure monitoring)
- http://stackshare.io/performance-monitoring
Top picks:
- Stackdriver - Free with alerting
- mackerel.io - Free for 5 server metrics
- Librato - Affordable collectd integrated monitoring
- OpsDash - Says self hosted but not sure if that's accurate. $1/host/mo
- BMC TrueSign Pulse - Free
Final decision:
We are going with Librato. Reasoning below:
Librato has better ranking on stackshare than Stackdriver. Mackerel has no stackshare page.
- http://stackshare.io/librato
- http://stackshare.io/stackdriver
- http://stackshare.io/opsdash
- http://stackshare.io/boundary (BMC TrueSight Pulse)
OpsDash doesn't inspire confidence due to emphasis on self-hosted.
Boundary/TrueSight Pulse has good companies but it's been acquired or rebranded so not certain any more.
Bonus: librato is used by companies we know/respect (e.g. Heroku, Travis CI).
All solutions:
- cron + cronitor.io or Dead Man's Snitch
- Super simplified monitoring solution that would do things like
df -h
to verify we have good disk space - We can pair this with a weekly
collectd
graph report via email - Dead Man: $5/mo for hourly checks, $19/mo for 15 min checks - https://deadmanssnitch.com/plans
- cronitor.io: $25/mo for basic - https://cronitor.io/pricing
- Super simplified monitoring solution that would do things like
- Nagios (self-hosted)
- Might be tedious to set up/maintain
- Datadog (SaaS)
- Free for no alerting
- $15/host/mo for alerting
- bigpanda - $449/mo nope
- CloudMonix - Free but no alerts; $75/mo for alerts
- Stackdriver - Free with alerting Winner?
- icinga - Self hosted
- monit - Self hosted
- monitis - $5/mo for basic server checks every 5 min
- CFEngine - Self hosted
- mackerel.io - Free for 5 server metrics Secondary choice?
- Munin - Self hosted
- Opsmatic - Private beta despite looking completed o_O
- sensu - Self hosted
- server density - $10/mo
- Zabbix - On-site enterprise?
- Zenoss - On-site enterprise?
- Librato - Affordable collectd integrated monitoring Third choice
- Kadira - Unsure what N/A hosts means ($50/mo otherwise)
- OpsDash - Says self hosted but not sure if that's accurate. $1/host/mo Fourth choice?
- Netuitive - $15/mo
- Stackify - $15/mo
- AppNeta TraceView - $79/mo
- BMC TrueSight Pulse - Free Fifth choice?
- SignalFX - $15/mo
- Sealion - Free for 1 server and 12 hour retention (valid choice but sounds awful)
- LogicMonitor - Requires a quote. Fuck off.
- vector - Self hosted
- Instrumental - $150/mo. Noooooope
- Ruxit - $144/mo ($0.20/host/hour)
- Inspeqtor - $25/mo
- Network Polygraph - 49EUR/mo