Analisis Availabilitas dan Reliabilitas Multi-Master Database Server Dengan State Snapshot Transfers (SST) Jenis Rsync Pada MariaDB Galera Cluster

Penulis

Mahendra Data, Gilang Ramadhan, Kasyful Amron

Abstrak

Abstrak

Sistem database merupakan bagian yang tak terpisahkan dari aplikasi berskala enterprise. Data didalamnya merupakan aset yang sangat penting, sehingga data tersebut tidak boleh rusak terlebih lagi hilang. Itulah sebabnya mengapa reliabilitas dan availabilitas sistem database menjadi hal yang sangat penting. Berbagai cara telah dikembangkan untuk meningkatkan reliabilitas dan availabilitas sebuah sistem database, salah satunya adalah teknik replikasi. MariaDB Galera Cluster adalah salah satu DBMS open source populer yang memiliki mekanisme replikasi. MariaDB Galera Cluster memiliki beberapa metode State Snapshot Transfer (SST) pada saat proses replikasi, yaitu rsync, mysqldump, xtrabackup, xtrabackup-v2. Kurangnya pemahaman administrator sistem terhadap perilaku tiap metode SST dapat mengakibatkan error pada sistem database. Untuk mencegahnya diperlukan analisis yang mendalam tentang dampak dan perilaku metode SST yang digunakan. Penelitian ini fokus pada analisis kinerja dari metode SST jenis rsync. Rsync dipilih karena metode ini merupakan metode SST default yang digunakan oleh MariaDB Galera Cluster. Dari hasil percobaan disimpulkan bahwa jumlah node dalam klaster menjadi hal yang perlu menjadi perhatian. Klaster yang hanya memiliki dua node akan sangat rentan terkena gangguan jika salah satu node terputus koneksinya atau mengalami down atau crash, sehingga jumlah node minimal dalam satu klaster yang disarankan adalah tiga node agar availabilitas dan reliabilitas MariaDB Galera Cluster dapat terjaga dengan baik.

Kata kunci: database, klaster, availabilitas, reliabilitas, MariaDB Galera Cluster


Abstract

The database system is an integral part of enterprise-scale applications. The data in it is a very important asset, so it may not be damaged or lost. That’s why the reliability and availability of the database system become very important. Various ways have been developed to improve the reliability and availability of a database system, one of them is a replication technique. MariaDB Galera Cluster is one of the popular open-source DBMS that has a replication mechanism. MariaDB Galera Cluster has several methods of doing Snapshot State Transfer (SST) during the replication process, namely rsync, mysqldump, xtrabackup, xtrabackup-v2. Lack of understanding by the system administrator of the behavior of every SST method can lead to errors in the database system. To prevent it, depth analysis of the impact and the behavior of the SST methods is required. This study focused on analyzing the performance of rsync SST method. Rsync is chosen because it is the default SST method used by MariaDB Galera Cluster. The experimental results show that the number of nodes in a cluster should be concerned. Clusters which only have two nodes would be highly vulnerable to disruption if one node disconnected or experience down or crash. We recommend the minimum number of nodes in a cluster is three so that the availability and reliability of MariaDB Galera Cluster can be properly maintained.

Keywords: database, cluster, availability, reliability, MariaDB Galera Cluster

Teks Lengkap:

PDF (English)

Referensi


ADITYA, B. & JUHANA, T., 2015. A high availability (HA) MariaDB Galera Cluster across data center with optimized WRR scheduling algorithm of LVS - TUN. In 2015 9th International Conference on Telecommunication Systems Services and Applications (TSSA). IEEE, hal. 1–5.

DOMASCHKA, J., HAUSER, C.B. & ERB, B., 2014. Reliability and Availability Properties of Distributed Database Systems. In 2014 IEEE 18th International Enterprise Distributed Object Computing Conference. IEEE, hal. 226–233.

MARIADB, 2014a. About Galera Replication. Tersedia di: https://mariadb.com/kb/en/mariadb/about-galera-replication/ [Diakses Januari 22, 2017].

MARIADB, 2014b. About MariaDB. Tersedia di: https://mariadb.org/about/ [Diakses Januari 12, 2017].

MARIADB, 2015. MariaDB Galera Cluster - Known Limitations. Tersedia di: https://mariadb.com/kb/en/mariadb/mariadb-galera-cluster-known-limitations/ [Diakses Januari 19, 2017].

MARIADB, 2017. MariaDB Galera Cluster Configuration Variables. Tersedia di: https://mariadb.com/kb/en/mariadb/galera-cluster-system-variables/.

PACITTI, E. ET AL., 2005. Preventive Replication in a Database Cluster. Distributed and Parallel Databases, 18(3), hal.223–251.

YURCHENKO, A., 2009. MySQL patches by Codership. Tersedia di: https://launchpad.net/codership-mysql [Diakses Januari 19, 2017].




DOI: http://dx.doi.org/10.25126/jtiik.201741288