Selecting a monitoring system is hard, but there are just four important questions which could simplify this work.
The first question to be answered when choosing a monitoring system is, “What problem do I want to solve with it?”
- I have a lot of servers, databases, and other equipment in the network, but I do not have a way of monitoring them of monitoring the overall system;
- I provide a service to my clients, but the only way of knowing about problems are when they are reported by the clients themselves;
- I try to monitor their IT infrastructure through the use of a lot of different specialized software that requires a lot of specialists to support;
- I have a monitoring system, but am unhappy with the cost / quality / functionality.
Clearly articulate the answer to this first question and you can move on.
The second question is, “What is my budget for this monitor?”
Today’s market offers a wide variety of the monitoring software ranging from free monitoring systems such as Zabbix to the mid-range products such as dxMARS, AccelOps PAM, on up to expensive monitoring systems such as IBM’s Tivoli Monitoring.
How does one calculate the final cost of the monitoring system?
The quantization of productivity in most modern monitoring systems is measured in the so-called Processing Unit (PU). This unit is the CPU core, or a device such as a switch, or a software module. The greater the number of Processing Units that you want to monitor, the greater will be the final price of the system. Also, the cost of the monitoring system should include the initial installation of the system, the training of staff, and the annual supporting fee. Note that even in “free systems” the training, implementation and supporting fee end up costing between 50,000 and 100,000 $.
Price is not always the unambiguous criterion of choice. Selection should be based on the specific tasks that you are going to solve and future needs of the monitoring system. It is not necessary to spend a million dollars if you only need to carry out monitoring of your corporate IT infrastructure.
Let’s examine the most important common problems:
- Tracking and analyzing vital parameters of the equipment (CPU usage, HD free space, temperature, etc.) of the data center in real time;
- Monitoring and analyzing the application’s parameters (the number of online customers, the daily transaction processing speed of applications, etc.) of the software (financial, telemetry, etc.) together with the system parameters for the equipment where it is installed;
- Predicting the occurrence of resource shortages and equipment failures in the geographically distributed IT infrastructure of the company;
- Reducing staff technical support services by means of reducing the amount of software used, and improving the overall availability of services to 99.9% on a 7/24/365 basis.
Each of the above problems can be partially or completely solved with one of the available monitoring solutions, but the difference will be the speed and quality of the achieved results. As the need for additional tasks are identified, such as keeping the history, special remote agents, the integration with third party systems and other features that can sometimes tip the balance in favor of a particular monitoring system.
The third question – How will I be able to implement monitoring and the training of employees?
Usually in more expensive monitoring solutions it takes more time and training to gain proficiency. The same can be said about the implementation. Also, in most cases the user will have to buy additional expensive server equipment and acquire annual certification for it. In return you will get almost 100% confidence that the solution will cover all of your needs. In the mid-market systems deployment takes less time and the hardware requirements are simply a minimum system configuration. Such a system straight out of the box will meet 80-90% of the needs of the customer, and after personalizing with modifications that are not usually too expensive, it may satisfy 100% of needs. In the case of a free monitoring system, the risk is poor performance and a cumbersome user interface, as often these systems are strung together from a large number of open source modules written by different people. Safety issues associated with public domain open source code has stopped many companies from the purchase and use of these, but new modules for such systems can be written independently. Conversely, expensive systems have problems related to the development of special features and white labeling, as required modules will be available in no less than six months to a year.
The last (fourth) question often only arises after the system has been implemented and the staff has been trained, but it should be considered in advance: What can be done in case of an error?
The answer is the same for monitoring systems in all price ranges—to contact support. In principle, free commercial support systems are not very different from the same service in expensive systems. The thing is to make sure that your contract guarantees that the service will be available to you 24/7, 365 days a year, otherwise you may find yourself in an awkward position when the monitoring system and your IT infrastructure are down and there is no one to call.
Any other questions? We will be glad to answer.