Assessment of the distance learning server\'s operation strategies and service capacity in advance
Download 399.25 Kb. Pdf ko'rish
|
e3sconf ebwff2023 06016
4 Methods
As mentioned above, the service time of the system is one of the most important factors for any web platform, and it plays an important role in attracting the attention of future users of the system. Therefore, one of today's major problems is selecting service devices for systems before integrating them into work processes and figuring out whether they are compatible with the flow of possible users. In this study, based on the special parameters (technical indicators) of the system server for distance education platforms, the calculation of its probable service time was carried out using available analytical models. First of all, the following technical specifications for the server for the system were chosen: HP Proliant DL 380 G6 processor: Intel Xeon E5540 (2.53 GHz, 8 cores), HDD 1.5 TB, RAM 22 GB. After determining the extra variables required to calculate the server's service time, the server processing time, queue time, CPU speed, RAM size, and storage capacity were calculated. The network delay (ND), which in this case depends on the user-server distance and Internet speed, was estimated in this study to be 200 ms on average. The average server processing time (SPT), which accounts for the time spent on database queries, computations, and processing by another server, was determined to be 100 ms. With regard to the number of requests on the server, server capacity, and numerous other considerations, the queue time (QPT) was also considered to be 30 ms. It should be noted that these times were averaged for E3S Web of Conferences 420, 06016 (2023) https://doi.org/10.1051/e3sconf/202342006016 EBWFF 2023 4 user and server distances of up to 5 km and that they could change depending on requests and other influencing factors such as the user and server's distance, the programming language used to create the program, and the database used in it. This value was used to determine the server response time (SRT) for a single request, which was as follows: SRT = ND + SPT + QPT = 200 ms + 100 ms + 30 ms = 330 ms (1) It should be noted that this SRT value was estimated with the assumption that the server would be managing the average workload for a web application and sending requests over the internet. However, depending on other influencing factors, this value might be different in real-life situations. Following that, the server's capacity for simultaneous users was calculated using the aforementioned indicators. The server's hard drive and RAM device sizes, SRT, and the number of processor cores all took on significant significance in this situation. First of all, the total processing power (TPT) of the system server was determined by multiplying the clock speed of each core of the Intel Xeon E5540 (2.53 GHz, 8 cores) processor by the number of cores. TPT = 2.53 GHz * 8 cores = 20.24 GHz (2) The following formula was used to compute the number of server requests served per second (TRPS) based on these identified indicators: TRPS = ்் ௌோ் ൌ ଶǤଶସ Ǥଷଷ௦ = 61.3 requests (3) The following step involved analyzing the service time of the server based on the number of user requests and the SRT that was previously calculated using the available models. The M/M/m model, which is based on the Poisson distribution of the probability of the reduction in user flow and assumes an exponential distribution of service times, was used in this procedure to execute the change in service time [20–22]. The service time based on this model was calculated in this instance using formula (3) [23–25]. RT = ቀ ଵ ఓ ቁ כ ͳ ቀ ሺכఘሻ మ ଵିఘ כ ሺ݉ െ ߩሻቁ (4) Here, μ – is the service speed, m – number of service devices, ρ – server utilization rate (%) (that is, the rate of arrival of requests λ divided by the rate of service of each server(μ*m), ρ = λ/(μ*m) in this, λ – speed of requests entering the system), and RT (Response Time) is average service time. In this instance, the TRPS value calculated above was deemed to be the service speed. And the number of devices m was taken at the expense of server cores. Based on this, it was determined that the level of server utilization ρ will have the following values based on various indicators of the number of requests entering the system (Fig. 1). Download 399.25 Kb. Do'stlaringiz bilan baham: |
Ma'lumotlar bazasi mualliflik huquqi bilan himoyalangan ©fayllar.org 2024
ma'muriyatiga murojaat qiling
ma'muriyatiga murojaat qiling