You were redirected from a different version of the documentation. Click here to go back.


A lightweight program that provides an interface between your MongoDB processes and Cloud Manager.
agent API key

A unique identifier that authenticates a project’s MongoDB Agents to Cloud Manager. A project can have multiple agent API keys.

authentication mechanism

A method to enable access to a MongoDB database. This is separate from authorization, which grants you permission to use certain actions on a MongoDB database.


The assisted management of MongoDB processes through the Cloud Manager interface. The MongoDB Agents installed on your MongoDB hosts allow you to deploy, configure, and update MongoDB processes directly from Cloud Manager.

See also


Automation Agent

The lightweight component that had automated common management tasks. The Automation Agent runs on every server that ran a mongod or mongos.


This has been replaced with the MongoDB Agent.

Backup Agent

The lightweight component that had run within your data center and backed up MongoDB processes via the MongoDB wire protocol. No direct file system access was needed.


This has been replaced with the MongoDB Agent.


The collection of jobs needed to restore a sharded cluster: one job for each shard and one job for the config server.

Replica set restore jobs do not use batches.


A point in time between snapshots to which you can restore a sharded cluster. Cloud Manager must stop the balancer each time it creates a checkpoint. Cloud Manager doesn’t require checkpoints and disables checkpoints by default.


You may use checkpoints for clusters that run MongoDB with Feature Compatibility Version of 4.0 or earlier. Checkpoints were removed from MongoDB instances with FCV of 4.2 or later.

See also


In Cloud Manager, cluster can refer to either a replica set or sharded cluster.
custom snapshot

A backup of the state of your MongoDB deployment at a point in time between stored snapshots. Cloud Manager builds a custom snapshot by applying oplog data to a stored snapshot.

See also

Restore Overview


A process that eliminates redundant data. This data can be additional copies of database or file system documents or redundant data within those documents at the block level when written to a storage medium like spinning or solid state disks. Only unique documents or blocks are written to a storage medium. This process usually applies to backups or data archiving.


Each recipient in an email system may have their own copy of an email sent to the whole company. With deduplication, all copies of this except one are replaced with pointers to a single stored copy before backing up the email system. This effectively reduces the amount of storage capacity needed to back up this one email by 99 percent.

Usually refers to all the MongoDB processes that run within a Cloud Manager project. Deployment can also refer to a specific set of MongoDB processes, such as a specific sharded cluster or replica set.
dirty bytes
Data that has been updated in the WiredTiger cache but not flushed to disk.
excluded namespace

A database or collection that Cloud Manager will not back up, as designated by its namespace.


A distinct set of MongoDB processes and Cloud Manager users. Synonymous with project.

See also


A physical machine, virtual machine, or container that serves one or more MongoDB processes.
initial sync

The MongoDB operation that replicates data from an existing replica set member to a new member. Cloud Manager uses initial sync when starting a new backup.

See also

Replica Set Data Synchronization in the MongoDB manual

A string that contains the information necessary to connect from Cloud Manager to Atlas during a Live Migration from an Cloud Manager deployment to a deployment in Atlas.

When you are ready to live migrate data from an Cloud Manager deployment, you generate a link-token in Atlas and then enter it in your Cloud Manager organization’s settings. You use the same link-token to migrate each deployment in your Cloud Manager organization sequentially, one at a time.

You can generate multiple link-tokens in Atlas. Use one unique link-token for each Cloud Manager organization.

migration host

A dedicated server with its own specially-configured MongoDB Agent configured for Live Migration. You run the Live Migration (push) process on the migration host to migrate your MongoDB deployment to Atlas.

MongoDB Agent

A lightweight agent that can monitor, manage, and back up your MongoDB databases.

See also

MongoDB Agent


The real-time reporting, visualization, and alerting of the state of your MongoDB processes.

See also


Monitoring Agent

The lightweight component that had run within your data center and monitored your MongoDB processes via the MongoDB wire protocol. No direct file system access was needed.


This has been replaced with the MongoDB Agent.


The combination of the database name and collection name:

oplog slice
Compressed batch of entries for the tailed oplog of a backed-up shard or replica set. The MongoDB Agent creates an oplog slice and sends it to Cloud Manager, which stores it in the Oplog Store Database.
Oplog Store Database
The database where Ops Manager stores oplog slices before applying them to a deployment’s backup.
A data transmission that the MongoDB Agent sends to Cloud Manager to confirm that the Agent and its MongoDB processes are running and reachable.
point-in-time restore

A database restoration that captures the state of your data at a moment in-between snapshots. Point-in-time restores take longer to perform than snapshot restores.

See also

Restore Overview


An instance of MongoDB running on a given host and port. The MongoDB database process is mongod. MongoDB also uses the mongos process to route operations in the sharded clusters.


A distinct set of MongoDB processes and Cloud Manager users. Synonymous with group.

See also


Public API key
A unique identifier that authenticates a Cloud Manager user through the Cloud Manager Administration API. The key belongs to the user, as opposed to the agent API key, which belongs to the project.
queryable backup

A feature provided by Cloud Manager in which Cloud Manager quickly and securely makes a given snapshot accessible over a MongoDB connection string. You can use the connection string with standard MongoDB tools such as mongosh or mongodump to access the snapshot for read-only operations.

Queryable backups start up quickly regardless of the snapshot’s total data size. They are uniquely useful for restoring a small subset of data, such as a document that was accidentally deleted, or reading out a single collection with mongodump.

Recovery Point Objective
The maximum tolerable age of backup files that must be recovered from storage for normal operations to resume after a failure or disaster occurs.
Recovery Time Objective
The maximum tolerable length of time that a system can be offline after a failure or disaster occurs.

The permissions granted to a Cloud Manager or MongoDB user.

See also

rolling restart

A technique used to maintain cluster availability during maintenance periods by updating nodes in a replica set one-by-one, always maintaining a primary node, until all nodes are updated.

A physical or virtual machine that serves one or more MongoDB processes.

Single backup of your data that Cloud Manager captures at a specific interval and stores. The Snapshot Frequency and Retention Policy determines the interval for taking snapshots and how long to store them. You can query specific backups.

See also

custom snapshot

snapshot frequency and retention policy

The schedule for how often to take snapshots and how long to store them.

snapshot store
The location where your snapshots are stored.
storage engine

The database storage engine manages how data is stored on disk. MongoDB versions 3.0 and later offer multiple storage engines.

See also