Hosted monitoring systems review: for easy to use monitoring without extra hassle.

Below is the list of the market key players (in the alphabetical order):

  • AppFirst
  • NewRelic
  • ScoutApp
  • ServerDensity
  • 10bees

 

AppFirst

This company is focused on providing with a comprehensive server monitoring and analysis tool.

They provide an agent (called ‘collector’), installing a kernel module to get as much information about the processes as possible. Linux, FreeBSD, Windows and Solaris support is available. They also support connecting other data sources like ‘statsd’.

The company is heavily working on the customized dashboard and has a lot of the technology behind the service.

Mobile app was not found, but it doesn’t seem to be that much needed. The solution comes with some alarms pre-defined and enabled by default. In case of any problems, an operating system status snapshot, with the top resources consuming process, is sent it to you via e-mail, which is quite handy.

Price: $15 or $25 per server per month, depending on the number of monitored servers.

 

NewRelic

Born as a programmer’s profiling and an application internal behaviour measurement tool, it was extended to support Linux, SmartOS and Windows operating systems monitoring.

The system has a powerful dashboard with the graphs and a correlations analysis. The installation procedure is a bit complicated (as opposed to the AppFirst and 10bees approach), but once installed, it’s managed by OS’ packaging system.

Mobile app is available, but there are no default alarming settings. The idea is that you have a single place to hold and work with your data – would it be a server, application or anything else.

Price: for the operating systems supported – it’s free.

 

ScoutApp

A Ruby based agent and server parts of the product enable Linux and a limited MacOS X and Solaris monitoring.

For me it seems to be too much Ruby oriented, the dashboard functionality looks to be quite limited. After adding the monitoring agent to the cron manually, my server disk was overflowed with the logs, which did not make me happier either.

Price: $10 (with 5 minutes time resolution delay) or $15 per server per month.

 

Server Density

A Python based product in both – agent and server parts. The system provides with a complete monitoring for Linux and Windows, and with a very limited FreeBSD and MacOS X support.

Originally started as a server monitoring service, it is getting more features with each release, moving to the direction of New Relic. You can also send your data from your apps via the API, and monitor your site via the http ping tool and the system even has some basic cloud based server management.

For me the dashboard did not seem good enough for the purpose of servers monitoring, but it is getting more and more flexible, so that you can adapt it according to your needs.

The other drawback is monitoring agent and it’s requirements – ServerDensity’s Python based agent is constantly a reason for the AppFirst monitoring to throw an alarm, saying that Python took all my micro EC2 instance CPU time. Windows version of the agent is .NET based and requires few hundred megabytes to run.

There is a nicely looking mobile app, with an iOS support introduced recently.

Price: $10 per server per month (comes with 1 web check as well).

 

10bees

A new player on the hosted system monitoring market, started a while ago and recently restarted with a new platform and solutions on the background.

10bees solution doesn’t have that much on the board yet, but it’s the only hosted monitoring system providing with a full support for *BSD world – FreeBSD, NetBSD and OpenBSD are supported, alongside with Linux, Windows and MacOS X.

The dashboard is straight forward, but a bit simplified, in comparison to the other players. The servers are represented like in a rack. History perspective is currently limited to one hour and no alarming feature as of now.

The agent is mostly written on Perl.

 

Price: free.

 

Summary

If your server resources are limited (in terms of CPU and memory) – consider using AppFirst or 10bees; if you don’t really care about the memory and CPU and Linux is your primary platform – then almost all of the solutions will work for you.

For support of the wide range of the operating systems, note AppFirst and 10bees solutions. If you need NetBSD or OpenBSD support – 10bees is your only choice.

In case you are a developer and more focused on the languages / development platforms, and don’t care that much on the operating system support – NewRelic or ServerDensity looks the best choice.

And if you search for a perfect one-suits-all blender for all kind of needs – NewRelic seems to be an absolute winner here, with an option to be integrated with almost anything, ready to pick up templates and a platforms support.

For the development, quality assurance and a build purposes, we have to have a good operating systems environment (a thing that we, internally, call a ‘lab’).

Currently we support NetBSD, OpenBSD, FreeBSD, Linux, Windows, MacOS X in 32 and 64 bit versions – all of them have to be always on available to make sure our Agent’s releases pass our internal Q&A tests and will be working properly in the real world.

To address this need, we had a few always available Xen hosted VDS servers, and a set of the virtualized machines, running on a usual desktops. That was fine at the beginning, but when we started growing and put more pressure on the Agents’ build quality, it turned out that we need something better than that.
We could have bought more from NQHost.com hosting – they proved to have a good technology behind the virtualization (so far that is a single known hosting company allowing for almost every modern operating system for a reasonable price), good support and even a so called ‘Low Cost VPS’. The problem is that those instances have a very good CPU resources (especially compared to Amazon’s low cost solution), but not that much RAM.
Other good and proven option is a CloudSigma, but this platform flexibility comes at a price, so we had to reject this option as well.

With a use of a simple calculator it turned out that moving all of the virtual machines to any hosted environment would be a killing cost; add the costs of maintaining – both the software and a hardware, and it’s turn out to be clearly a no-go decision.

So, the question is: how to get relatively cheap virtualized containers, without a need to maintain a physical server, operating system and learning a new technology?

Our solution turns out to be a combination of Hetzner’s ‘Dedicated server’ (which is especially attractive through the auctioning service) and a VMWare ESXi. Hetzner support team is taking care about the hardware. VMWare ESXi – is a turnkey solution for virtualization, free to use (if you can live with the restrictions).
We’ve got a server from the auction and in a few working hours, I already had an IP to connect to a web based Hetzner console. In a few minutes, I opened an ESXi console (needs Windows 7 or higher, no Linux, nor web versions, unless you’ve get a paid license) and started rolling out our new virtual infrastructure.

As a result, we have got something like 20 virtualized servers running without extra hassle and for a very reasonable price, providing a good stable environment for the development of a new features, support new operating systems releases and covering our quality requirements with an automated constantly running tests.

We just released a new version of our service, and we are proud to highlight the major changes.

On the web site side:

  • we addressed a few cases when the dashboard was not working due to some user’s local network settings;
  • history graph logic improved, it is now better handling interruptions and big servers configuration;
  • small ‘i’ sign next to the server name provides short information on the OS and the agent version;
  • few Internet Explorer related fixes;
  • improvements in site navigation.

On the agent side:

  • fixed some bugs on CentOS 5.* platform;
  • brought back a support for NetBSD and OpenBSD platforms;
  • introduced a FreeBSD 10 pkgng support;
  • our user verified and found 10bees solution working on FreeNAS 9.

Infrastructure:

  • reorganized for a better latency between the front/back-end application and the database;
  • setup a better internal operating systems testing environment;
  • significantly improved our centralized collector for a better handling of the requests and optimized the performance.

We have also introduced a ZenDesk support platform, to make sure all of the requested are served in the best way.

Please, let us know on your wishes / features request, or any other comments via support@10bees.com, or right here on the comments.

You haven’t heard from us for a while and there was a reason for that. Since the initial release a while ago we’ve been working hard on all of the feedback we received from the community to make things right.

We had to rework the system in quite a few major aspects, and internally it is very different from what it was originally, which gives us more room to grow.

In a new release we seriously increased the dashboard performance, fixed quite a few bugs, added some minor nice features, and we decided to remove the new features, that were not working great: a deep history overview, NetBSD and OpenBSD agents.

We will keep you posted on our plans and a new features we are working at, and we expect to polish NetBSD and OpenBSD agents, as well as provide a seriously rewritten and PKGNG ready FreeBSD 10 port.

Follow

Get every new post delivered to your Inbox.