[Yandex Cloud documentation](../../index.md) > [Yandex Managed Service for YDB](../index.md) > [Concepts](index.md) > Terms and definitions

# Terms and definitions for Managed Service for YDB

## Database {#database}

You can create a [YDB](https://ydb.tech/docs/en//concepts/databases#database) database in Managed Service for YDB in a serverless configuration or with [dedicated servers](../operations/manage-databases.md#create-db-serverless). For more information about differences between configurations, see [Serverless and dedicated modes](serverless-and-dedicated.md).

### Serverless database {#serverless}

Computing resources for serverless YDB DBs in Managed Service for YDB are provided automatically for running DB queries.

The payment amount depends on the CPU capacity actually used and the I/O operations required to run a database query. The amount of data stored in the database is paid additionally.

### Database on dedicated resources {#resource-presets}

A Managed Service for YDB DB is provided on VMs running on platforms supported by Yandex Compute Cloud. For more information, see [Platforms](../../compute/concepts/vm-platforms.md).

{% note warning %}

For reliable and stable performance, a database needs multiple slots. A database run in the production environment must have at least three slots.

{% endnote %}

In Managed Service for YDB, the following VM configurations are available:

| Platform | Configuration name | Number of vCPUs | Guaranteed vCPU share | RAM, GB |
| ----- | ----- | ----- | ----- | ----- |
| **Intel Cascade Lake** | medium | 8 | 100% | 32 |
| **Intel Cascade Lake** | medium-m64 | 8 | 100% | 64 |
| **Intel Cascade Lake** | medium-m96 | 8 | 100% | 96 |
| **Intel Cascade Lake** | large | 12 | 100% | 48 |
| **Intel Cascade Lake** | xlarge | 16 | 100% | 64 |
| **Intel Cascade Lake** | olap-c16-m128 | 16 | 100% | 128 |
| **Intel Cascade Lake** | oltp-c16-m128 | 16 | 100% | 128 |

Databases run on allocated computing resources are paid on an hourly basis. The amount of data storage for the database is paid additionally (see [Storage groups](#storage-groups) below).

## Cluster {#cluster}

A YDB cluster is a set of YDB nodes the load is distributed across.

The Managed Service for YDB team takes care of cluster maintenance, while application developers can work directly with YDB.

For more information about YDB clusters, see the [documentation](https://ydb.tech/docs/en//concepts/databases#cluster).

## Storage groups {#storage-groups}

Yandex Managed Service for YDB uses three disks in each of the three availability zones, both for databases on dedicated resources and serverless databases. This configuration ensures fault tolerance if a zone or disk fails, as well as provides a redundancy factor of 3.

For more information about storage groups in YDB, see the [documentation](https://ydb.tech/docs/en//concepts/glossary#storage-group).

## Temporary storage (spilling) {#spilling}

Yandex Managed Service for YDB OLAP databases use **spilling**, a memory management mechanism to save temporary computation data to the disk if there is not enough RAM. Spilling enables user queries that involve processing massive amounts of data, when temporary data exceeds the available node memory.

For more information about spilling in YDB, see [this guide](https://ydb.tech/docs/en//concepts/spilling).

## Regions and availability zones {#regions-az}

Managed Service for YDB databases are hosted in three availability zones of the central Russia `ru-central1` region. Managed Service for YDB ensures the databases are fully available if any availability zone fails.

You can learn more about Yandex Cloud availability zones in [Availability zones](../../overview/concepts/geo-scope.md). To move a database to a different availability zone, follow [this guide](../operations/migration-to-an-availability-zone.md).