Martin Rusev here, the guy behind Amon. I just want to provide a little bit more information about where the project is headed. First I want to clarify that Amon is not a new project. It has been in development for 2 years now, the SaaS version is in Beta, but I consider the underlying technology pretty stable and mature. My big goal with Amon is to move beyond monitoring into server analytics and recovery. That means - to analyze the data and send meaningful, troubleshoot friendly alerts that show the whole server state, not just "your cpu usage is > 70%". On recovery - The idea behind is to automate simple sysadmin tasks, like cleaning up log files, restarting a process, etc. As an example - if you have an alert: mysql is down for 5 minutes, execute: service mysql start
Hi Martin, I love the simplicity of Amon (and those worried about installing via a curl script, I'm sure he can give you manual directions). We don't really need app profiling like New Relic provides and we have a lot of clusters running, so the affordability is something we like also. Being able to know that our cluster load is getting high and we should deploy more via a link to our Ansible server would be a very welcome feature.
Sorry, but those are terrible examples. If your processes keep crashing, you should find and fix the underlying problem, or if you're forced to use unstable software, rely on something like upstart or daemontools to keep it running. And if your apache logs regularly fill your disk, add them to logrotate.
Sorry about the examples, the general idea is that you can execute a command on the server when there are alerts for a specific problem. Upstart and daemontools can't keep your processes running. For this specific task there are tools like Monit, God, Puppet to some extend
Perhaps a better example would be automating a previously manual failover, i.e. if a service becomes unavailable on one machine, start it on the warm standby. Or in EC2, terminate an instance and start a new one.
Promising project but the New relic comparison is misguided. New Relic does have server monitoring but it's core strength is in application layer with agents for java, ruby, .net, python, php applications.
I came here to say the same thing. NewRelic is actually pretty bad at saying "your server is down". But, it's incredible at saying "Your landing page is loading slowly because the time it takes to load your data from MySQL is slow. Oh, and here's the offending query." Different things.
Adding in the ability to mark when a deploy goes out makes it really easy to see when deploys caused issues historically (and immediately for those not involved with the deploy itself).
Indeed. Their application metrics have been incredibly useful for us at Pathwright. It's nice always knowing what views are slow, which comprise the bulk of our CPU time. The SQL time breakdowns are handy to.
You can, of course, do all of these things on your own without New Relic, but a small team like us needs to be spending time on other things (like pushing the product forward).
There's also Scalyr (https://www.scalyr.com), built by two ex-Google devops engineers. (Full disclosure, I work for them.)
There's no shortage of options out there. What we've heard from people is that while New Relic is powerful, it has a steeper learning curve and can be cumbersome for small/medium companies. I suspect that's why many alternatives are popping up.
Also their pricing.. The company I work for has no problem paying but my side projects where I'd like the same level of visiblity.. I just can't justify the cost
Thanks for the link on Scalyr, will definitely check it out. Btw, it wouldnt hurt to throw up some pictures on your site :)
Thanks for the feedback! :) Many others seem to agree with you about the graphics, so we're definitely working that into our re-design, which is in progress now. I hope you try us out anyway!
This is equivalent to the server monitoring of the New Relic platform, but not the application monitoring side of things, which I think is what most people pay New Relic for.
If you just want quick & cheap server monitoring -aaS, there's http://www.serverdensity.com/, with a pretty straightforward Python plugin system for your custom metrics.
None of this compares to what New Relic gets you with their deep code introspection. Server monitoring is just an additional gimmick you get with it, mostly because its nice to have all your metrics in one place when you hunt down a problem (there's a CloudWatch forwarder than fetches you all manner of AWS metrics into NR too).
I tried Server Density and was really depressed by my experience.
During my first couple of hours I encountered a ton of bugs, all of which I reported to SD. Some of them were UI issues — I would edit a chart's settings or an alert, and a minute later my changes had been reverted to the original one. Some of them were monitoring issues, with alerts showing up 45 minutes after the service went down (I was testing with a non-production setup), or not at all, or showing false positives and/or wrong values. And a good deal of SD was not well designed, eg. no monitoring templating.
Also sad to see that plugins cost money; SD comes with very little functionality built in, so you have to rely on plugins.
In general, the whole system seemed incredibly brittle. Not something I would trust a production system to. They have been running SD since 2009, but it feels like an alpha version; not a good sign. (That said, SD's response was very forthcoming.)
I switched to Scout [1], which is not as visually pretty, but works and seems very sturdy. No bugs found so far. Integrates very well with Puppet, easy plugin development in languages like Ruby. Oh, and plugins are free.
+1. I've used Cloudkick (before Rackspace), Copperegg (not too bad either) and New Relic. Server Density is VERY reasonably priced, simple and gets the job of server monitoring done. The custom metrics are key. New Relic for example doesn't support custom metrics. A custom metric example: my Sidekiq queue sizes. When Sidekiq first came out, there was no way to alert/monitor this. It only took a few hours of Python.
What they are lacking in comparison to running your own NMS (e.g. NetXMS or Zabbix) is making custom views (the Dashboard feature in NR is pretty useless).
If you liked Cloudkick I think you should also check out https://mist.io. In fact we started mist.io when Cloudkick was about to close down. We like to think of it as mobile friendly Cloudkick with a twist.
Thanks for the shout-out! I'm on the https://www.scalyr.com team and am here if anyone has questions/feedback/suggestions. It grew out of our own frustration with the available tools, exactly as you express.
(Scalyr founder here) Thanks for the feedback. Agreed, we haven't done much to present the product yet. For the last couple of years, we've been focusing on building out the product while keeping our early users happy -- making sure we can walk the walk before we talk the talk. We'll be launching a completely new home page in a month or so that should do a much better job of presenting what we do. In the meantime, if anyone out there is frustrated with the current monitoring offerings, drop us a line (I'm steve@$COMPANY.com). We have a lot to offer, and we'll go the extra mile and beyond on support.
Thanks for that feedback. We're redesigning the homepage now and will include more screenshots. You can check out the live demo in the meantime to see it in action (the app UI will also get a facelift soon): http://goo.gl/BZ5GNd
The difference between the hosted version and the open source version is that the latter can only monitor a single server - you'd need to run a copy of the open source version for each server you have.[0]
I don't want to take the spotlight away from Amon here, it looks like a good product and the idea of focusing on meaningful analysis is nice (instead of only storing droves of raw stats).
I don't like the install script you use. First of all you use sudo hardcoded in the script, it makes me nervous about what you are going to do on my server. I know it's nonsense because a deb can do about anything you want, but it would make me feel better.
Lots of people mention New Relic application monitoring - One of the alternatives that I've created is App Enlight - https://appenlight.com - giving you application level visibility
it's worth pointing out that new relic does not charge for server monitoring (SmartOS, *Nix, Windows). It's a free product and doesn't matter if you have one server or 100,000 or where they're hosted.
