Network Daemon Software Development For Agricultural Cloud Platforms

In the wake of developments in smart agriculture and cloud computing,the scale of the agricultural cloud platform has also become huge,the resources on the cloud platform are colorful and dynamically distributed without geographical restrictions.So with minimal costs and smallest influence to ensure agricultural cloud platform of high efficiency,stable operation is worth studying.Therefore,it is very meaningful to develop the daemon software of agricultural cloud platform.Currently,the common open source daemons include Cacti,Nagios,Ganglia,while they all have their own defects.Cacti performs other functions only by manual installation of plug-ins except drawing function;Nagios can only alarm in real time and has no historical data;Ganglia neither has built-in message notification system nor alarm mechanism.Most obviously,none of them have a network traffic warning mechanism.This paper develops a more easy-to-use,extensible,high-availability and early-warning system of daemon software through the analysis of the architecture and advantages of the commonly used daemons.The main research contents are as follows: Combining push and pull algorithm to optimize the collection of agent;Realizing load balancing through consistent hashing algorithm;With the help of keepalived to make the service highly available;Carrying out the high concurrency and disaster recovery backup of daemon data by using MySQL master from replication;The short-term prediction of traffic flow is implemented based on the time series ARIMA traffic prediction model,and measures can be taken in advance to avoid the occurrence of network congestion.Finally,testing CPU load,memory utilization rate,network card import and export flow to verify the reliability of software functions;the push-pull algorithm shows advantages in data acquisition through comparative tests in push-pull algorithm and push algorithm;the software’s performance has been greatly improved in procing data and the read performance of the MySQL database by comparison testing.Which also proves the feasibility of applying ARIMA model to traffic prediction through MATLAB simulation experiment.