Reduction of the harmful effect of critical modes in the operation queue environment for authorization protocols for large requests

Abstract

An essential part of web security is keeping the payload intact from changes. The data during transmission could be changed, where
the encryption is not used, or the data gets decrypted in the middle of the transmission. In our previous researches, the “chunking”
method was introduced, which was compared with the “Buffering to file” method. The comparison showed the reduction of recourse
consumption. In a multithreading environment, to manage resources efficiently, it is vital to distribute the workload among processor
cores. A decent solution for using multithreading efficiently is operation queues. However, if too many operations are accumulated in
the operation queue, the system falls into the critical mode. It is characterized by the increase of memory consumption, which may
cause the instability of the system. In the course of the study, the main parameters were determined, influencing the data processing
speed, and insignificant ones were excluded from the calculation. Earlier, a method was developed for determining the conditions for
the falling of a system to a critical mode. It was used as a starting point for the experimental research. A new method based on the
method of identifying critical modes in the operation queue is proposed. It differs from existing ones by the ability to simulate critical
modes at a given workload, which allows predicting critical modes in order to reduce their negative effect. A series of experiments
were carried out, and the results were used to study the dependences of memory consumption on the number of connections and
writing speed in critical modes. From the study, three types of critical modes were determined. This made it possible to establish the
patterns of the emergence of critical modes in information systems and their impact on the available memory. The formulas are
obtained that approximate the experimental data: the dependence of the used memory on the number of connections and the write
speed. The research results can be used in the development of information systems and the analysis of failures