Server monitoring: The Foundation for IT performance

Complete monitoring for any type of server

Monitoring your servers is a business-critical task. And still, most companies face disruptions in server performance or even permanent damage. Slow and unstable performance of IT infrastructure leads to inefficiency, unsatisfied customers and a loss in revenue.

The main reasons are blind spots in your server monitoring or the fact that important assets had to be excluded from the monitoring system for other reasons. Without visibility you cannot gain insights. This lack of detail thus leads to issues that cannot be localized and mitigated.

Physical servers need monitoring just like virtual servers do. The latter also includes cloud assets and virtualized servers. You can very quickly end up with many different types of servers that are set up for various purposes.

Depending on the environment, there are several sources for gathering monitoring information that you should pay attention to. Windows, Linux and other operating systems provide output for monitoring. Vendors have their own interfaces for monitoring bare-metal-servers. IPMI and SNMP can also be used to monitor servers.

In the case of virtualized servers, you can get data from the hypervisor and other sources as well. If you are working together with a cloud service provider, you can use their APIs to extract monitoring information and get it into your own monitoring platform.

The main challenge in server monitoring is balancing all of these different monitoring task, and having suitable approaches for individual scenarios. Checkmk is the all-in-on monitoring platform, and it provides agents for all established operating systems (windows, linux and others), but it is also able to monitor virtualized servers (such as Microsoft Hyper-V or VMware ESX), and servers in cloud environments (for example AWS or Microsoft Azure).

Checkmk discovers your hardware & software automatically, and you can use it for centralized asset management to track changes in your inventory. And with the Raw Edition you can do all that for free.

Get started: The basics of server monitoring

Now, which values and functions should you track? It doesn't matter whether you monitor a virtual web server or a physical Windows server, these four areas have to be part of the monitoring of any server:

• CPU: Is there enough capacity for all operation, and is the computing power efficiently distributed?
• RAM: Do you have enough memory for all applications and the cache?
• Block or object storage: Is there enough storage on the system, and does the data match the expected throughput?
• Network: Who has access to the data on the server? At what speed is the data transported to which areas?

This is just the beginning. The challenge is to grasp the complex interactions of your server landscape. Most infrastructure today is not dependent on physical assets. There are more virtual servers and assets hosted in the cloud.

In addition, container solutions like Docker or Kubernetes allow flexible infrastructure deployment without requiring a central management. This is indispensable for dynamic IT processes, but also comes with a new challenge for monitoring.

Checkmk can monitor all resources in one platform: It can monitor highly agile dynamic environments, but also static assets. The approach is not limited to servers only, but can also monitor network devices, data centers and other assets. This way you can analyse issues in depth.

Finding the right monitoring tool

A first area to consider is the technical maturity of a product. It has to be able to gather detailed information from a host while using as few resources as possible. It also should not interfere with the operations of the server, but rather use passive mechanisms.

Checkmk combines status-based, metric-based and log-based data in one tool, so that you can create warnings and alerts for every server with minimum effort.

From a strategic perspective a tool should be open to users without prior monitoring experience – this is the only way to avoid silos. Checkmk provides a comprehensive handbook, video tutorials and trainings.

Checkmk additionally comes with its own graphing, dashboarding and reporting engines to create metrics and analyses for different roles. It also integrates with external visualization tools like Grafana. This way you can make the best use of the data – no matter if you are a monitoring veteran or work in a totally different department.

And last, but not least, you should consider the total costs of ownership that go beyond the costs for software licenses: An IT department needs to spend time on administration and management of the monitoring – and should not be overwhelmed with false alerts in day-to-day operations.

Tools with a low usability or high user requirements lead to more expenses as more budget has to be spent on hiring matching talent. There is also a similar risk in case a tool is not able to be scaled. The costs for additional hardware, service providers and consultants can get out of hand if your monitoring is not optimized for larger environments. Checkmk has a highly efficient architecture, and it requires minimal hardware resources. It is easy to implement and can be operated with minimal manpower – even in very large environments.

Expert insights into the daily best practices

The decision for which data you should monitor depends on the exact setup. If you deploy your servers onsite you have to keep an eye on the hardware of course. If you run a virtualized environment, you still have to monitor the hardware if it is provided by yourself.

On top of that you need to keep an eye on the hypervisor and the virtual machines (VMs) as well. If you run cloud servers you may shift some responsibilities to a service provider, but – or maybe especially because of that – you should still monitor these assets. In this way you can check on different service-level agreements (SLA). And ultimately it makes sense to monitor your spending on a service provider too.

Here is a practical example to show the importance of having data in depth.

Imagine monitoring a server on which the value for the CPU load stays above the total number of available cores for a longer period. The server is underperforming. But at the same time the CPU usage is not even close to 100 percent, so you probably have an issue within the server.

If you only monitor the CPU utilization, you would not be able to see any issues. If you just focus on the CPU load, you will get an alert, but probably falsely increase the number of cores. It is likely that many tasks block the cores, but the tasks do not seem to get processed on time. With Checkmk you have a rich source of data and could have a closer look at the throughput of the storage or the RAM and could investigate why things are being delayed. The in-depth information helps companies in making better decisions.

Agents, SNMP or IPMI: Pros & cons of different approaches

The basic architecture of the monitoring makes a big difference, because it not only decides on the quality of the information you get, but also on the level of security and the system requirements of your monitoring.

Because IT environments can change quickly, a platform should not be limited by its design. If companies adapt new technology their monitoring should adapt as well.

Over the years Checkmk has developed an extensive set of high-performance integrations (1,800+ plug-ins) for various operating systems. Because of our experience we suggest using agent-based monitoring as the first go-to approach for servers.

The agents require minimal CPU and RAM on the host. Additionally, the workload on the monitoring server itself is kept to an absolute minimum. Its footprint in the network is also smaller compared to other technologies.

Using Checkmk agents is very secure, too, because the connection does not need to make an active call. On the host the agent works in read-only-mode and supports end-to-end-encryption.

While there are benefits that speak in favor of using agents, Checkmk also works well with other sources of information. If using software on the host is not an option, you can use the native integration of Checkmk with SNMP or IPMI.

In the daily routine our agents have proven their ability, especially if combined with our agent management tool, the Agent Bakery. Especially in larger server environments many assets can be handled and adjusted easily and without writing permissions for the hosts. Malicious code injections are impossible thanks to our security-by-design approach. Compared to monitoring via SNMP, WMI or other approaches, the requirements for resources on the host and on the Checkmk server will always be smaller if you rely on our agents.

How to set up a holistic server monitoring within minutes

Getting your monitoring started with Checkmk only takes a couple of minutes. The Checkmk instance itself is deployed as a Linux server. After the deployment, you are good to go. There is no need for additional software or another database. All editions are available on the Checkmk homepage. You can also run Checkmk as a Docker image, and you can use Checkmk as a physical or virtual appliance. So you don’t need to have any experience with Linux to use Checkmk.

After the installation you can work with the graphical interface or the command line to interact with Checkmk. Nearly any host can be integrated into the monitoring, and there are several ways to do so – including auto-detection mechanisms for integrating large numbers of hosts or dynamic environments automatically into Checkmk. Aspects of each host will be attached in the form of services.

If you use agents to monitor your servers, they will connect to TCP port 6556. Only on receiving a Checkmk server query will they be activated and then respond with the required data. Check intervals can be as short as one second. If you are using SNMP Checkmk will send a UDP package (Port 161), and receives an answer in the same format.

Agents and ports can be adapted, and Checkmk is able to use any input as long as it conforms with the guidelines for coding check plug-ins. The easiest way to extend agents is with local checks. Alongside the more than 1,800 official plug-ins, you can find extensions written by our Checkmk community members; most of these are available in the Checkmk Exchange.

Linux server monitoring

Using Linux as the operating system for servers is a common practice. There are several distributions laid-out especially for the needs of servers. Examples are Red Hat Enterprise Linux, Fedora, CentOS, openSUSE, SLES, Debian, Ubuntu – but there are many more. This is no surprise as Linux is known for running very stably and efficiently.

Checkmk can monitor any Linux server because our agent consists of a simple shell script that routes data to the TCP port 6556.

Checkmk picks up the information and gets it into the monitoring. The agent is not compiled, and all functions are fully transparent. Even if this sounds simple, knowhow is important when it comes monitoring Linux servers.

For example, Linux has its own way of managing its memory, and you should consider that in your monitoring. We have trusted in Linux for a long time and are happy to share our experience. Follow the link below if you want to read more about Linux server monitoring.