Experimental study of the automatic troubleshooting system in the database

Modern applications are focused on cloud services to achieve better performance, geographic replication and lower cost of ownership. Following the modern concepts of cloud services, this study draws on rich telemetry data and displays the workload performed using the Azure SQL database. The main purpose of this research is the potential improvement both of service and customer assistance employing a controlled platform. The automatic database troubleshooting system is designed to detect problems in a relational cloud database and analyze appropriately the sources of problems in order to reduce the time and cost of manual search and solution of these problems. This system was implemented on top of the Microsoft Azure platform. It is based on scientific models of general and categorical statistical data, which were developed and constructed after a thorough examination of the collected telemetry data. The final root cause of each current issue in the Azure service is gathered after analyzing the results of the models by means of an expert system. The evaluation results show that the continuous enhancement of the infrastructure has reduced the processing time, approximately, by 2 times while the number of intervals has doubled, which can be considered an overall improvement of 4 times, approximately.

1. Danilov A.D., Sinyukov D.S. The mechanism of data distribution on special transactions with real-time operational content based on caching in heterogeneous objects of a distributed network. Informatsionnyye tekhnologii modelirovaniya i upravleniya = Information technologies of modeling and control. 2021;125(3):216–223. (In Russ.)

2. Danilov A.D., Sinyukov D.S. An approach to transaction management in heterogeneous distributed replicated database systems in real time. Sistemy upravleniya i informatsionnyye tekhnologii = Control systems and information technologies. 2021;85(3):59–65. (In Russ.)

3. Sinyukov D.S., Danilov A.D. Application of database management systems as a service in complex information systems. Trudy Vserossiyskoy nauchnoy konferentsii «Dostizheniya nauki i tekhnologiy-DNiT-2021» = Proceedings of the All-Russian Scientific Conference "Achievements of Science and Technology-DNiT-2021". Krasnoyarsk; 2021. Available from: http://ru-conf.domnit.ru/media/filer_public/42/d3/42d35c66-d78a-411c-b74d-9ef2f99ff7b6/3006-dnit-2021.pdf. (In Russ.)

4. Sinyukov D.S. Problems of troubleshooting in databases. Modern informatization problems in simulation and social technologies (MIP-2022’SCT): Proceedings of the XXVII-th International Open Science Conference. Yelm, WA, USA. 2022;162–174.

5. Jeyakumar V., Madani O., Parandehgheibi A., Kulshreshtha A., Zeng W., Yadav N. ExplainIt! – A declarative root-cause analysis engine for time series data. SIGMOD'19 2019 International Conference on Management of Data, Amsterdam, Holland. 2019;333–348.

6. Raeder T., Dalessandro B., Provost F. Design principles of massive, robust prediction systems. 18th ACM SIGKDD international conference on Knowledge discovery and data mining, Beijing, China. 2012;1357-1365.

7. Automatic Performance Diagnostics. Oracle; 2017. Available from: https://docs.oracle.com/database/121/TGDBA/pfgrf_diag.htm#TGDBA026.

8. Automatic SQL tuning. Oracle; 2018. Available from: https://docs.oracle.com/cd/B28359_01/server.111/b28274/sql_tune.htm #CHDJDFGE.

9. Oleinikova S.A., Kravets O.Ja., Aksenov I.A., Frantisek O.Yu., Rahman P.A., Atlasov I.V. The general scheme of the genetic algorithm for solving the task scheduling problem for a multistage system and assigning time for jobs. International Journal on Information Technologies and Security. 2021;13(4):47–58.

10. Mustafayev V.A., Zeynalabdiyeva I.S., Kravets O.Ja. Control model of parallel functioning production modules as fuzzy Petri nets. Journal of Physics: Conference Series. 2021;2094:022003. Available from: https://doi.org/10.1088/1742-6596/2094/2/022003.