Product Description Managed Databases

Managed Databases is a service for deploying and managing high-performance and fault-tolerant clusters of supported databases in the cloud.

You can work with Managed Databases in the Control panel, via the Managed Databases API, or with Terraform (except for OpenSearch).

The product supports user types and roles, projects, and project limits and quotas.

Records of operations with Managed Databases are saved in audit logs.

Supported Managed Databases

PostgreSQL	An open-source database. Focused on performance and extensibility — you can connect any external data sources, create new data types, and functions
PostgreSQL for 1C	A version of PostgreSQL with necessary extensions for efficient work with 1C:Enterprise
PostgreSQL TimescaleDB	A version of PostgreSQL with the TimescaleDB extension, which can be used for storing time-series data
PostgreSQL PGVector	A PostgreSQL version with the pgvector extension that can be used for working with vector data and semantic search
MySQL semi-sync	An open-source relational database management system, easy to manage and scale. Suitable for most data operations
MySQL sync	An open-source MySQL solution based on Percona Server for MySQL with the XtraDB storage engine
Redis	A NoSQL in-memory database management system. Can be used as a database, queue system, message broker, and for data caching. Works with key-value data
Kafka	An open-source distributed system for message delivery, storage, and processing. Can act as a data bus for Cloud Native applications
OpenSearch	A scalable open-source search and analytics system
ClickHouse®	A column-oriented SQL DBMS for real-time online analytical processing (OLAP)

How Managed Databases work

Managed databases are deployed in a cluster. A cluster consists of one or more database servers (nodes). Cluster nodes run on the basis of Cloud Platform resources.

Managed Databases support cluster monitoring, backups, and scaling. You can increase the fault tolerance of the cluster and configure replication between nodes.

Database settings when creating a cluster are set by default and depend on the cluster configuration and database version. You can change them if necessary.

The configuration of managed database networks depends on the specifics of the infrastructure into which the managed database is integrated.

Monitoring

In Managed Databases, you can monitor the cluster state in the control panel:

view cluster node usage and database load information as charts in the control panel;
view the cluster status;
receive disk space notifications.

Cluster node and database metrics can also be exported in Prometheus format.

For more information about monitoring, see instructions for PostgreSQL, PostgreSQL for 1C, PostgreSQL TimescaleDB, PostgreSQL PGVector, MySQL sync, MySQL semi-sync, Redis, and Kafka.

Backups

In Managed Databases, backups are created automatically for all supported DBMSs, except Kafka and ClickHouse®. The backup creation mechanism and frequency depend on the selected DBMS.

Backups are stored in Selectel S3 isolated from backups of other users. The backup retention period is 7 days.

You can restore a cluster or data from backups. Backups cannot be downloaded. Automatic backup creation cannot be disabled.

Learn more about backups in the instructions for PostgreSQL, PostgreSQL for 1C, PostgreSQL TimescaleDB, PostgreSQL PGVector, MySQL sync, MySQL semi-sync, Redis, OpenSearch.

Scaling

A Managed Database cluster can be scaled — for example, you can increase vCPUs and RAM to improve cluster performance. You can select a configuration from another configuration line, but the disk type must match and the disk size must be larger.

Learn more about scaling in the instructions for PostgreSQL, PostgreSQL for 1C, PostgreSQL TimescaleDB, PostgreSQL PGVector, MySQL sync, MySQL semi-sync, Redis, Kafka and OpenSearch.

Fault tolerance and replication

PostgreSQL, PostgreSQL for 1C, PostgreSQL TimescaleDB, PostgreSQL PGVector, MySQL sync, My SQL semi-sync, Redis

By default, a cluster consists of one primary node — the master node. When connecting to the master node, all operations are available: read (SELECT) and write (INSERT, UPDATE, DELETE, and others).

To ensure cluster fault tolerance, add replicas — full copies of the master node. They are available for reading data (SELECT) only. If the master node is unavailable, the replicas will take over its role, and the cluster will continue to operate normally. They can also be used to reduce the load on the master node during heavy read operations.

For a cluster with replicas, an SLA applies — we guarantee 99.95% write availability and 99.99% read availability.

The node placement type in a cluster depends on the presence of replicas in the cluster, the type of pool where the cluster is located, and the number of segments in the pool.

Learn more about fault tolerance in the instructions for PostgreSQL, PostgreSQL for 1C, PostgreSQL TimescaleDB, PostgreSQL PGVector, MySQL sync, MySQL semi-sync, Redis.

OpenSearch

The fault tolerance of an OpenSearch Managed Database cluster is affected by:

index replication;
the number of nodes in node groups;
the node placement type in node groups.

Learn more in the OpenSearch cluster fault tolerance instructions.

Managed Database settings

Database settings affect database cluster performance. When creating a database cluster, values for all settings are configured automatically. The values are selected to ensure high cluster performance and vary depending on the cluster configuration and database version.

If the automatic values do not fit your needs, for all Managed Databases except Redis and OpenSearch, you can set custom values when creating a cluster or change settings in an already created cluster.

Learn more about Managed Database settings in the instructions for PostgreSQL, PostgreSQL for 1C, PostgreSQL TimescaleDB, PostgreSQL PGVector, MySQL sync, MySQL semi-sync and Kafka.

Networks

When creating a Managed Database cluster, you need to consider the specifics of the infrastructure into which the cloud database is integrated — whether access to cluster nodes from the internet is needed and whether network connectivity with other Selectel services is required.

A cluster can be connected to:

a private subnet — a subnet without internet access;
a public subnet (except OpenSearch and ClickHouse®) — all addresses of the public subnet are accessible from the internet.

Once the cluster is created, the subnet cannot be changed.

Learn more about setting up network connectivity between a Selectel dedicated server and a Managed Database cluster in the instructions for PostgreSQL, PostgreSQL for 1C, PostgreSQL TimescaleDB, PostgreSQL PGVector, MySQL sync, MySQL semi-sync, Redis and Kafka.

Areas of responsibility

Selectel provides

hardware selection for high DBMS performance;
operating system installation;
DBMS installation and standard configuration that is optimal for the selected configuration;
operating system and utility software updates and maintenance;
cluster reliability and fault tolerance — when you create a fault-tolerant cluster, we provide automatic failover in case of failure;
configuration and maintenance of the utility network for cluster replicas;
backups — automatic backup creation and storage;
the cluster health monitoring system in the control panel;
secure data storage and protection against theft and leaks;
compliance with 152-FZ requirements;
resource availability for cluster scaling if you initiate scaling;
technical support.

The user provides

correct connection to the database;
optimal database query design;
data schema and structure in the database;
changing settings according to their workload profile and application specifics;
initiating cluster scaling.

If you need help with database administration, order managed services.

ClickHouse® is a registered trademark of ClickHouse, Inc. https://clickhouse.com.