E.A. Eliseeva1, B.S. Goryachkin2, M.V. Vinogradova3, M.V. Chernen'kij4

1–3 Bauman Moscow State Technical University (Moscow, Russia)
 

This study evaluates the execution time of queries to clustered databases: Neo4J, MongoDB, Cassandra, PostgreSQL. The experiment was carried out on the Ubuntu operating system running on a virtual box on windows with a database cluster deployed on several servers, where the master server performed operations to write data and transferred the generated changes to other servers, and the replicas (slave) read data from database, optimizing the load between servers.

The experimental database contained three entities: user (User), user rights (UserPermit) and scientific articles (ScientificArticles). Entities were filled with strings ranging from 1 to 300,000 values. To determine the query execution time, two types of data insertion request (simple and complex), as well as two types of data reading requests (simple and complex) were selected.

In query No. 1 (simple insert) with a small amount of data (up to 1000 rows), the Cassandra DBMS turned out to be the most productive. When adding large amounts of data (more than 1000 rows), the MongoDB DBMS showed the best results.
In query No. 2 (simple reading), up to 1000 rows, the MongoDB DBMS won, and when reading more than 1000 rows of data, the PostgreSQL DBMS took the lead.

In query #3 (complex insert), Cassandra and MongoDB showed similar results to query #1. However, unlike a simple insert, a complex one added not one row of data to the User entity, but four with a repeating user_permit field when executing one request.

In query No. 4 (complex reading), data was selected for all fields with queries to the User, UserPermit and ScientificArticles tables and users were selected by address and then all access rights and scientific articles were obtained from such users. As a result, with a large sample, it turned out to be difficult to identify a clear leader, but when the number of rows was reduced to 100, the Cassandra DBMS showed the best query time.

Based on requests for complex reading and inserting data, analytical dependencies were formed, which were also experimentally verified and confirmed with an error of 0,16 to 32%.

Goryachkin B.S., Vinogradova M.V., Eliseeva E.A., Chernen'kij M.V. Estimating search execution time in NoSQL and object-relational databases. Dynamics of complex systems. 2022. V. 16. № 2. P. 44−51. DOI: 10.18127/j19997493-202202-05 (in Russian).

1. Goryachkin B.S. Ergonomicheskij analiz sistem obrabotki informacii i upravleniya. Vestnik evrazijskoj nauki. 2017. T. 9. № 3. C. 72 (in Russian).
2. Orlov S.A. Tekhnologii razrabotki programmnogo obespecheniya. SPb.: Piter. 2002 (in Russian).
3. Vinogradova M.V., Belousova V.I. Unificirovannyj process razrabotki programmnogo obespecheniya. M.: MGTU im. N.E. Baumana. 2015 (in Russian).
4. Vinogradova M.V., Anisimov N.A. Ustanovka i nastrojka klastera Hadoop na baze kontejnernoj virtualizacii LXC. Inzhenernyj vestnik. 2015. № 4 (in Russian).
5. Grigor'ev Yu.A., Plutenko A.D., Pluzhnikova O.Yu. Relyacionnye bazy dannyh i sistemy NoSQL. Amurskij gos. un-t. 2018 (in Russian).
6. Grigor'ev Yu.A. Ocenka vremeni vypolneniya SQL-zaprosov k bazam dannyh. Mashinostroenie i komp'yuternye tekhnologib. 2012. № 01 (in Russian)
7. Grigor'ev Yu.A., Ermakov E.Yu. Analiz vremeni vypolneniya zaprosa v parallel'nom kolonochnom hranilishche dannyh. Inzhenernyj zhurnal: nauka i innovacii. 2013. № 11 (23) (in Russian).
8. Ermakov E.Yu. Matematicheskaya model' vremeni vypolneniya zaprosa v parallel'nom kolonochnom hranilishche dannyh i primer ee ispol'zovaniya. Sovremennye informacionnye tekhnologii i IT-obrazovanie. 2013. № 9 (in Russian).
9. Grigor'ev Yu.A., Ermakov E.Yu. Sravnenie processov obrabotki zaprosa k odnoj tablice v parallel'noj strochnoj i kolonochnoj sisteme baz dannyh. Inzhenernyj zhurnal: nauka i innovacii. 2012. № 3 (3) (in Russian).
10. Ermakov E.Yu. Matematicheskaya model' vremeni vypolneniya zaprosa v parallel'nom kolonochnom hranilishche dannyh i primer ee ispol'zovaniya. Sovremennye informacionnye tekhnologii i IT-obrazovanie. 2013. № 9 (in Russian).
11. Grigor'ev Yu.A., Pluzhnikov V.L. Algoritm vybora arhitektury parallel'noj sistemy baz dannyh po kriteriyu stoimosti. Mashinostroenie i komp'yuternye tekhnologii. 2011. № 12 (in Russian).
12. Danilov A.M., Gar'kina I.A. Interpolyaciya, approksimaciya, optimizaciya: analiz i sintez slozhnyh sistem. Penza: PGUAS. 2014 (in Russian).
13. Onlajn kal'kulyatory PLANETCALC. URL: https://planetcalc.ru/5992/? (data obrashcheniya: 01.04.2022) (in Russian).