Resource relationships in Yandex MPP Analytics for PostgreSQL

The main entity used in Yandex MPP Analytics for PostgreSQL is a cluster.

A database cluster consists of hosts, which are virtual machines with databases deployed on them. Learn more about databases in Yandex MPP Analytics for PostgreSQL.

A database cluster includes:

• Two master hosts (primary and standby).
• Two or more segment hosts.

The primary master host (PRIMARY) accepts client connections and SQL queries from applications using the Odissey connection pooler and distributes them to the segment hosts for processing.

The standby master host (STANDBY) continuously replicates the primary master host's data but accepts no user connections.

If the primary master host fails, the standby one takes over its functions. This way, a cluster with two master hosts will continue processing queries if any single master fails.

Segment hosts store distributed cluster data and process queries.

Segment host architectureSegment host architecture

The segment host architecture is presented on the diagram:

Segment hosts have standalone database management systems, known as segments, deployed on them. The number of segments is the same for each host. Each cluster segment has a replica, i.e., a mirror segment located on another host, which stores a copy of the main segment's data.

Each segment includes the following components:

• Execution engine.
• Configurable extensions.

You can store data in a cluster storage or hybrid storage.

You can expand a cluster by adding more segment hosts. When expanding a cluster, you can also increase the number of segments per host. You cannot increase the number of segments without expanding a cluster.

Executing queries in a clusterExecuting queries in a cluster

Whatever your use case, the application sends an SQL query to the cluster. The cluster processes the query as part of the user session as follows:

1. Establishes a session with the master host with the help of the Odyssey connection pooler, which manages authorization and reduces cluster overheads.
2. Forwards the query to the Query Coordinator process on the master host. Query Coordinator analyzes the query, breaks it into parts, and selects an execution plan for each one using the Optimizer.
3. Forwards the query parts with execution plans to the segments. When executing a query, the segments can access a local storage as well as external data sources with the help of Yandex Data Transfer or external tables.

The resource relationships in a Yandex MPP Analytics for PostgreSQL cluster and the data exchange model within the cluster, as well as between the primary master host and applications, are shown in the diagram:

Hosting cluster hostsHosting cluster hosts

Yandex MPP Analytics for PostgreSQL cluster hosts are cloud VMs. Such VMs can reside on:

• Regular Yandex Cloud hosts:

  These are physical servers for hosting cluster VMs. They are randomly selected from a pool of available hosts that meet the selected cluster configuration.

• Dedicated Yandex Cloud hosts:

  These are physical servers reserved exclusively for your VMs. VMs on dedicated hosts have all the features of regular VMs. In addition, they are physically isolated from other users' VMs and have access to the whole volume of the physical server's local disks.

  Dedicated hosts are selected from dedicated host groups specified when creating a cluster. You must first create a group of dedicated hosts in Yandex Compute Cloud.

  For more information, see Dedicated host.

All Yandex MPP Analytics for PostgreSQL cluster hosts reside in one of these availability zones: ru-central1-a, ru-central1-b, or ru-central1-d.

When creating a cluster, specify:

• Host class: Template for deploying cluster hosts. For a list of available host classes and their characteristics, see Host classes.

• Environment: Environment where the cluster will be deployed:

  • PRODUCTION: For stable versions of your applications.
  • PRESTABLE: For testing purposes. The prestable environment is similar to the production environment and likewise covered by an SLA, but it is the first to get new features, improvements, and bug fixes. In the prestable environment, you can test service updates for compatibility with your application.

Be mindful of what is managed by the service and what, by the Yandex Cloud customer. Understanding these areas of management helps you use your cloud resources efficiently and avoid potential database-related issues. For more information, see Zones of control between managed database (MDB) service users and Yandex Cloud.