3.1. Description of Components #
The Postgres Pro Enterprise Manager architecture comprises the components below.
- Web Application #
Web application provides a graphical user interface accessible through a browser. The web application is installed on the same server as the manager. Users perform various actions using the browser, and the web application transforms these actions into requests and sends them to the manager. The manager processes requests and returns them to the web application. The web application transforms the response into various representations and then displays it in the browser.
- Manager #
PPEM manager (hereinafter referred to as the manager) is a service that runs in the background. It accepts requests for DBMS infrastructure maintenance and has its own scheduler for performing regular service tasks. The manager also accepts requests from the web application and transforms them into operations according to the designed logic.
To perform operations, you need the following:
various service data, which is usually (but not necessarily) stored in the repository
executing instructions on the agent side
The manager reports the operation results to the web application upon completion.
Note
The operations performed by the manager can be synchronous or asynchronous. In case of synchronous operations, the manager is forced to wait until the operation completes to obtain the response. In case of asynchronous operations, the manager immediately receives a response that the operation has been queued for execution. In this case, the manager needs to periodically check the operation status, but, in most cases, the user receives a notification upon the operation completion.
- Repository #
PPEM repository (hereinafter referred to as the repository) is a database on a dedicated DBMS instance with a set of schemas and tables where the manager stores service data necessary for operation. When the manager starts, it establishes a connection with the repository and reads and writes data from and to it during operations. Repository availability is critical, since the manager cannot continue operating if the repository is unavailable.
- Agents #
PPEM agents (hereinafter referred to as agents) are services that run on managed DBMS servers. They receive control instructions from the manager and, depending on the instruction type, perform various actions both in the operating system and in the DBMS instance, for example:
running commands
creating directories and files
executing queries
During or after execution, the execution result is sent to the manager to be recorded in the repository and/or subsequently processed according to the operation logic.
Agents also have an internal service task scheduler that regularly performs background tasks and sends their results to the manager.
One agent is enough on a standalone server to serve one or more DBMS instances.
- DBMS Instance #
DBMS instance is a PPEM-managed object, i.e., the Postgres Pro DBMS in various editions (Postgres Pro Standard, Postgres Pro Enterprise). Several DBMS instances can be combined into clusters, typically streaming replication clusters.
Only a single agent interacts with a single DBMS instance. Avoid scenarios where multiple agents interact with the same DBMS instance, as this can lead to confusion and ambiguous behavior. From the DBMS instance standpoint, the agent is a regular application software that connects to the instance via the SQL interface using a predetermined DBMS account.
- External Services #
PPEM can use various external services to extend its features and capabilities. All these services are optional, and interaction with them is configured separately.
Note
PPEM does not include tools for administrative management of external services (e.g., resource and configuration control).
The following external services are supported:
LDAP directory is a directory of users and groups for authenticating users in PPEM. PPEM supports OpenLDAP and Microsoft Active Directory.
S3-compatible object storage is used by PPEM to store backups made with the pg_probackup utility.
OTLP-compatible metrics storage system is a metrics storage system that supports the push model with the ability to record metrics via the OTLP protocol, for example, Victoriametrics, and/or the pull model capable of fetching metrics via the HTTP protocol, for example, Prometheus. PPEM can use such a system to receive metrics that are written there by the pgpro-otel-collector. PPEM supports both Victoriametrics and Prometheus.
OTLP-compatible log storage system is a metrics storage system capable of receiving logs via the OTLP protocol. PPEM can use such a system to receive DBMS logs that are written there by pgpro-otel-collector. PPEM supports Elasticsearch.