Today, we are highly dependent on the technology but we must ensure that we monitor the devices and technology we use to run our businesses and ecommerce. In this regard, synthetic monitoring is highly important. The following points will help you understand the importance and process of synthetic monitoring.
Database Servers
These systems suffer very strongly from the activity of the users of the transactional applications. In addition to the classical monitoring of systems i.e. CPU, Memory and Disk, it is very important not to forget to monitor the other critical parameters such as the following;
Delays in Disk Accesses
The delays in disk accesses must be very low to guarantee the locking times in accesses to the minimum data that prevent certain requests from being delayed in an important way. For example, the systems based on Oracle are designed to work in an optimal manner with write-to-disk delays of less than 10 ms.
Relevant Parameters of BBDD
We need to analyze in detail the most relevant parameters regarding the synthetic monitoring of the metrics of these systems, so that we can advance to take into account the parameters such as Buffer Cache Hit Ratio, which indicates the percentage of the requests that are result of access to the pages that are in memory, or Average Lock Wait Time, which indicates the average time that requests have to wait to be executed for the reason that are critical to make sure the optimal service performance.
Disk Cabins
With the intensive use of virtualization, the level of activity that these systems suffer is very high. In addition, the fact that several different systems are often accessing the same set of the disks can create cross-penalties in the very difficult to detect ports. In this case, the problems in one application can cause performance problems in another application that have nothing to do with the design of the same or the servers on which they are running. It is, therefore, very important to monitor both the IOPS levels, number of the disk access operations per second that we are asking the booth in each given interval as the individual number of IOPS generated by each system. If some of them are generating in a prolonged way a very high number of accesses, it is necessary to investigate what is the cause.
We believe that taking these points into account when defining how synthetic monitoring of the systems should be done in the multilayered web architectures, as it is crucial to make sure the correct functioning of the systems.
