MongoDB

Architecture Guide
The foundational concepts that underpin the
architecture of MongoDB
MongoDB
Architecture Guide

Introduction
Data and software are at the heart of ° Previously separate transactional,
every business. But for many organizations, analytical, search, and mobile
realizing the full potential of the digital workloads are converging to create
economy remains a significant challenge. rich data-driven applications and
Since the inception of MongoDB, we’ve customer experiences. However,
understood that the biggest challenges each workload has traditionally
developers face are related to working been powered by its own database,
with data: creating duplicated data silos
stitched together with fragile ETL
° Demands for higher productivity pipelines, accessed by different
and faster time to market are developer APIs.
being held back by rigid relational
data models that are mismatched To address some of these challenges, non-
to modern code and impose tabular (sometimes called NoSQL or non-
complex interdependencies among relational) databases have been rapidly
engineering teams. adopted over the past decade. But many of
° Organizations are unable to work these NoSQL databases are simply Band-
with, or extract insights from, the Aids, offering a niche set of functionality.
massive and rapidly growing amount
The problem is that typical NoSQL
of data generated by modern
databases do one or two things well. They
applications, including time series,
might offer more flexibility in the data
geospatial, and polymorphic data.
model than traditional databases or scale
° Monolithic and fragile legacy out easily. But to do this, they discard
databases are inhibiting the
the most valuable features of relational
wholesale shift to distributed systems
databases. They often sacrifice data
and cloud computing that deliver
integrity and the ability to work with data in
the resilience and scale demanded
the ways needed to build rich and valuable
by digital business and support new
regulatory demands for data privacy. applications — whether these are new
digital touchpoints with an organization’s
customers, or modernized core back-end
business processes.

2
MongoDB
Architecture Guide

The Document Model

MongoDB was launched in 2009 as a
completely new class of general-purpose
database and quickly established itself
Improved support for
as one of the most popular databases time series data
among developers. MongoDB retains the
best aspects of relational and NoSQL Time series data is data that represents
databases while providing a technology how a system, a process, or a behavior
foundation that enables organizations changes over time. It may be captured
to meet the demands of modern at constant time intervals, like a device
applications. It does this by replacing measurement per second, or at irregular
the rigid tables of relational databases time intervals, as with alerts and event
with flexible documents that map to the audits. Time series data is critical for
way developers think and code. Instead modern applications, in particular for
of storing data in columns and rows, IoT, stock trading, clickstreams, and
document databases can store data as social media. With the move from
JSON (JavaScript Object Notation). A batch to real-time systems, the efficient
document database can store any type of capture and analysis of time series
data, and the structure of documents can data enables organizations to better
be easily modified. This enables developers detect and respond to events ahead of
to be far more productive and build or their competitors, improve operational
iterate upon their applications faster. You efficiency, and reduce cost and risk.
can add new fields without affecting other Thanks to MongoDB’s flexibility, teams
documents in the collection, the MongoDB have been using the database to store
equivalent of a table in a relational time series data for years. However,
database. And you can model data in any correctly modeling the data to achieve
way that suits the application, for example, a performant solution was not always
as key-value pairs, as the edges or nodes straightforward. With time series
of a graph, or as nested structures that collections — a new collection type
represent relationships. introduced with MongoDB 5.0 — and
new features such as clustered indexing
and window functions, teams can
work with and store time series data
without having to worry about low-
level model optimization. MongoDB will
automatically optimize your schema
for high-storage efficiency, low-latency
queries, and real-time analytics against
temporal data.

3
MongoDB
Architecture Guide

As developers have experienced the

“phone”: [
benefits of the document data model
{ “location”: “work”,
for themselves, it has become the most
“number”: “+44-1234567890”},
popular alternative to the tabular model
],
used by traditional relational databases.
“dob”: ISODate(“1977-04-01T05:00:00Z”),
The three primary advantages of the “retirement_fund”:
document data model are: NumberDecimal(“1292815.75”)
}

1. Intuitive: faster and

easier for developers With the document data model, there
is no need to decompose data across
Documents in the database directly map to tables, run expensive JOINs, or integrate
the objects in your code, so they are much a separate Object Relational Mapping
more natural to work with. (ORM) layer. Data that is accessed
The following example of a JSON document together is typically stored together, so
in MongoDB demonstrates how a customer you have less code to write and your users
object is modeled in a single document get higher performance.
structure with related data embedded as
subdocuments and arrays. This approach
collapses what would otherwise be seven
separate parent-child tables linked by
foreign keys in a relational database.

{
“The most beautiful
“_id”: part is the data model.
ObjectId(“5ad88534e3632e1a35a58d00”),
Everything is a natural
“name”: {
“first”: “John”, JSON document. So for
“last”: “Doe” },
the developers, it is easy
“address”: [
{ “location”: “work”, — really easy — for them
“address”: { to work quickly. They’re
“street”: “16 Hatfields”,
“city”: “London”, spending time on building
“postal_code”: “SE1 8DJ”}, business value rather than
“geo”: { “type”: “Point”, “coord”: [
51.5065752,-0.109081]}}, data modeling.”
], — Filip Dadgar, IT manager,
Toyota Material Handling Europe

4
MongoDB
Architecture Guide

2. Flexible schema: you can apply data governance

standards to a document schema while
dynamically adapt maintaining the benefits of a flexible data
to change model in development.

A document’s schema is dynamic and 3. Universal: JSON

self-describing, so you don’t need to
predefine it in the database. Fields can documents are
vary from document to document, and everywhere
you can modify the structure at any time,
allowing you to continuously integrate new Lightweight and language-independent,
application functionality without dealing JSON has become an established
with disruptive schema migrations. standard for data communication and
When you need to make changes to the storage. Documents allow you to structure
data model, the document database data in any way your application needs
continues to store the updated objects — rich objects, key-value pairs, tables,
without the need to perform costly “ALTER geospatial and time series data, and the
TABLE” operations, update a separate ORM nodes and edges of a graph. As a result
middleware layer, and coordinate all of of these properties, you can serve
these changes across multiple developer, many more classes of application with a
DBA, and ops teams. Documents allow single database.
multiple versions of the same schema to MongoDB stores data as JSON documents
exist in the same table space. Old and new in a binary representation called BSON
applications can coexist. (Binary JSON). Unlike most databases that
MongoDB also offers schema validation store JSON data as primitive strings and
so you can enforce rules governing the numbers, the BSON encoding extends the
structure of your documents. This is useful JSON representation to include additional
as your applications move into production types such as int, long, date, floating point,
because you can govern your schema and decimal128. This makes it much easier
without having to write controls in the for applications using MongoDB to reliably
application layer. With schema validation, process, sort, and compare data.

5
MongoDB
Architecture Guide

Working With Document Data

A few key differences among databases you’re manipulating data in a single
are the expressivity of the query document or with MongoDB’s scale-out
language, the richness of indexing, and architecture, across multiple documents, and
the data-integrity controls. geographically distributed in multiple shards.
The MongoDB Query API is comprehensive With strong data consistency, MongoDB
and expressive. Ad hoc queries, indexing, eliminates the application complexity
and real-time aggregations provide imposed by eventually consistent NoSQL
powerful ways to access, group, transform, systems. MongoDB’s consistency guarantees
and analyze data. You can federate queries are fully tunable, enabling you to balance
across databases, supporting transactional data freshness against performance.
workloads and archived data in your data
To make it easy for businesses to act on
lake using the same query API and drivers,
data in real time, many developers are
all with a single connection string.
building fully reactive event-driven data
The MongoDB Aggregation Pipeline pipelines. MongoDB goes beyond many
allows you to transform and analyze data. other databases with features such as
Documents enter a multistage pipeline Change Streams, which automatically
that transforms them into an aggregated detects and notifies consuming applications
result. The most basic pipeline stages of any data modifications in the database.
provide filters that operate like queries and
The MongoDB Query API absolves
document transformations that modify the
developers from having to research,
form of the output document. Other pipeline
learn, and stay up-to-date on multiple
operations provide tools for grouping and
ways to work with data across different
sorting documents by specific fields, as
workloads. It’s more natural to use than
well as tools for aggregating the contents
SQL because it feels like an extension of
of arrays, including arrays of documents. In
the programming languages developers
addition, pipeline stages can use operators
are already using. To further accelerate
for tasks such as calculating an average
developer productivity, MongoDB provides
or concatenating a string. The pipeline
native drivers for popular programming
provides efficient data aggregation using
languages and frameworks. Supported
native operations within MongoDB and is
drivers include Java, JavaScript, C#/.NET,
the preferred method for data aggregation
Go, Python, PHP, Scala, Rust, and more.
in MongoDB.
All supported MongoDB drivers are
With ACID transactions, you can maintain designed to be idiomatic for the given
the same all-or-nothing and snapshot programming language. This eliminates
isolation guarantees as with relational the need for cumbersome and fragile
databases. This remains possible whether ORM abstraction layers.

6
MongoDB
Architecture Guide

Distributed Architecture:
Scalable, Resilient, and
Mission Critical
Through replica sets and native sharding, identify those replica set members
MongoDB enables you to scale out your that have applied the most recent
applications with always-on availability. updates from the primary replica
You can distribute data for low-latency user
° Heartbeat and connectivity status
access while enforcing data sovereignty with the majority of other replica
controls for data privacy regulations such set members
as GDPR.
° User-defined priorities assigned to
replica set members
Availability and data By extending data protection,
protection with replica sets developers can configure replica sets
to provide tunable, multinode durability
MongoDB replica sets enable you to create and geographic awareness. Through
up to 50 copies of your data, which can be MongoDB’s write concern, you can ensure
provisioned across separate nodes, data write operations propagate to a majority
centers, and geographic regions. of replicas in a cluster. With MongoDB 5.0,
the default durability guarantee has been
Replica sets are predominantly designed
elevated to the majority (w:majority) write
for resilience. If a primary node suffers an
concern. Write success will now only be
outage or is taken down for maintenance,
acknowledged in the application once it
the MongoDB cluster will automatically
has been committed and persisted to disk
elect a replacement in a few seconds,
on a majority of replicas.
switching over client connections and
retrying any failed operations for you. Choosing the new default versus the former
w:1 default allows for a stronger durability
The replica set election process is
guarantee, where acknowledged data can
controlled by sophisticated algorithms
survive replica set elections and complete
based on an extended implementation
node failures. The new w:majority default
of the Raft consensus protocol. Before
setting is fully tunable, so you can maintain
a secondary replica is promoted, the
the earlier w:1 default or any custom write
election algorithms evaluate a range of
concern you had previously configured.
parameters including:
You can also create custom write concerns
° Analysis of election identifiers, time that target specific members of a replica
stamps, and journal persistence to

7
MongoDB
Architecture Guide

set, deployed locally and in remote regions. routing queries to a copy of the data
This ensures writes are only acknowledged that is physically closest to the user. With
once custom policies have been fulfilled, sophisticated policies such as hedged
such as writing to at least a primary and reads, the cluster will automatically route
replica in one region and at least one queries to the two closest nodes (measured
replica in a second region. This reduces the by ping distance), returning results from the
risk of data loss in the event of a complete fastest replica. This helps minimize queries
regional failure. waiting on a node that might otherwise be
busy, reducing 95th and 99th percentile
Beyond resilience, replica sets can also be
read latency. Note that hedged reads are
used to scale read operations, intelligently
available in shared clusters only.

Scale Up, Out, and

Across Storage Tiers
Like most databases, you can scale change your shard key — which determines
MongoDB vertically by moving to larger how data is distributed across a sharded
or smaller instance sizes. As a distributed cluster — on demand without impacting
system, MongoDB can perform a rolling system availability. As your shard key is
restart of the replica set, enabling you to modified or as you change the cluster
move between different instances without topology, MongoDB will automatically
application downtime. rebalance data across shards as needed
without manual intervention.
Through native sharding, MongoDB can
also scale out your database across By simply hashing a primary key value,
multiple nodes to handle write-intensive many distributed databases randomly
workloads and growing data sizes. Sharding spray data across a cluster of nodes,
with MongoDB allows you to seamlessly imposing performance penalties when data
scale the database as your applications is queried or adding application complexity
grow beyond the hardware limits of a when you need to locate data in a specific
single server, and it does so without adding region. By exposing multiple sharding
complexity to the application. policies to developers, MongoDB offers a
better approach. Data can be distributed
To respond to evolving workload demands,
according to query patterns or data
you can add and remove shards anytime.
placement requirements, giving you much
You also have the flexibility to refine or

8
 MongoDB
Architecture Guide

higher scalability across a more diverse set ° Zoned sharding: This allows
of workloads. MongoDB native sharding developers to define specific rules
gives you the following options: governing data placement in a
sharded cluster.
° Ranged sharding: Documents are
partitioned across shards according Beyond vertical and horizontal scaling,
to the shard key value. Documents MongoDB also offers tiered scaling. When
with shard key values close to one working in the cloud, the MongoDB Atlas
another are likely to be co-located Online Archive will automatically tier aged
on the same shard. This approach data out of the database and into cloud
is well suited for applications that
object storage. Archived data remains
need to optimize range-based
fully accessible with federated queries that
queries, such as co-locating data for
span both object and database storage in
customers in a specific region on a
a single connection string. This approach
specific set of shards.
enables you to more economically scale
° Hashed sharding: Documents are data storage by moving it to a lower-cost
distributed according to an MD5 hash
storage tier without losing access to the
of the shard key value. This approach
data and without grappling with slow and
guarantees a uniform distribution of
complex ETL pipelines.
writes across shards, which is often
optimal for ingesting streams of time
series and event data.

Figure 1: Serving always-on, globally distributed, write-everywhere apps with MongoDB Atlas Global Clusters

9
MongoDB
Architecture Guide

Privacy and Security

With the growing digital economy comes With client-side field-level encryption
an increase in governmental oversight (FLE), you have access to some of the
of privacy and data security. MongoDB most advanced data protection controls
includes extensive capabilities to defend, anywhere. FLE makes it even safer to
detect, and control access to data: store your most sensitive data in the cloud
because it’s completely inaccessible to
° Authentication: MongoDB offers
anyone who doesn’t have the encryption
a strong challenge-response
keys, including those running the database
mechanism based on SCRAM-256,
for you.
along with integration to enterprise
security infrastructure, including FLE also makes it easier to comply with
LDAP, Windows Active Directory, “right to be forgotten” conditions in
Kerberos, x.509 certificates, and privacy regulations, such as the GDPR and
AWS IAM. the CCPA. Simply destroy the customer
° Authorization: Role-based access key and the associated personal data is
control (RBAC) enables you to rendered useless.
configure granular permissions for
With FLE, you can selectively encrypt
a user or application based on the
individual document fields, each optionally
privileges they need to do their jobs.
secured with its own key and decrypted
° Auditing: For regulatory compliance, seamlessly on the client. In MongoDB, FLE
security administrators can use is totally separated from the database,
MongoDB’s native audit log to record
making it transparent to the server. Instead,
all database activity and changes.
it’s handled exclusively within the MongoDB
° Network isolation: For users drivers on the client. All encrypted fields
running fully managed databases on the server — stored in memory, in system
in MongoDB Atlas, user data logs, at rest, and in backups — are rendered
and underlying systems are fully as ciphertext, making them unreadable
isolated from other users. Database
to any party that does not have both
resources are associated with a
client access and the keys necessary to
user group, which is contained in
decrypt the data. This is a different and
its own virtual private cloud (VPC).
more comprehensive approach than the
Access can be granted only by IP
column encryption used in many relational
whitelisting or VPC peering.
databases. Most of these databases
° Encryption everywhere: MongoDB handle encryption server-side, so data is
data can be encrypted while in
still accessible to administrators who have
motion across the network, while in
access to the database instance itself, even
use in the database, and while at
if they have no client access privileges.
rest, whether on disk or in backups.

10
MongoDB
Architecture Guide

One Platform for All

Your Workloads
Building on MongoDB’s document At its core, MongoDB Atlas provides a
data model, expressive Query API, and general-purpose database (Atlas Database)
distributed systems DNA, the MongoDB for modern applications. Nearly every
Atlas application data platform delivers a application needs a fast database that can
cohesive and integrated set of data and deliver single-digit millisecond response
database services. Atlas streamlines how times. And with its flexible document
teams work with data, specifically in the data model, transactional guarantees,
context of building software and systems rich and expressive query API, and native
that deliver real-time experiences to both support for both vertical and horizontal
end customers and internal users. Key scaling, Atlas Database can be employed
characteristics of MongoDB’s application for practically any use case, reducing the
data platform include: need for specialized databases even as
requirements change. Cluster auto-scale
° A data plane with the ability to adjusts both compute and storage in
support a wide variety of application
response to application load, eliminating
types that can be independently
the need to monitor utilization and react to
developed, deployed, and evolved to
scaling needs.
address a wide variety of application
types and use cases Atlas Database is available in more
than 80 regions across AWS, Google
° A unified and consistent experience
for developers, data analysts, data Cloud Platform, and Azure. Best-in-class
scientists, and critical supporting infrastructure and database automation
functions such as operations teams, ensure continuous availability, elastic
security teams, and data engineers scalability, and compliance with the most
demanding data security and privacy
° Global, multi-cloud data
distribution — built on MongoDB’s standards. Uptime is backed by a 99.995%
native sharding — to support data service-level agreement.
residency requirements and provide
deployment flexibility
° Transparent data movement
between services and automated
data life-cycle management

11
 MongoDB
Architecture Guide

Beyond offering fully managed MongoDB

databases in the cloud, Atlas provides
additional complementary services
that allow organizations to support a
wide range of application and real-time
analytics data workloads.

Document Model

Key Value
Geospatial
Pairs

Relationships Graphs

Objects

Unified Interface Real-Time

Transactional Analytics

Search Mobile

Distributed
Security Architecture

Multi-Cloud

Figure 2: MongoDB’s application data platform

12
MongoDB
Architecture Guide

Real-Time Analytics
Atlas Database allows you to deploy a
The best way to run
read-only analytics node to serve more MongoDB in the cloud
resource-intensive analytics queries.
You can easily target analytics nodes by Atlas Database delivers MongoDB
configuring the read preference, effectively as a pay-as-you-go service billed on
ensuring that analytics queries leveraging an hourly basis. To deploy it, you can
MongoDB’s built-in aggregation pipeline use a GUI or the admin API to select
never contend for database resources the public cloud provider, region,
with your operational workloads. Analytics instance size, and features you need.
nodes, like all read-only nodes within a Atlas Database provides:
MongoDB cluster, do not participate in
° Automated database and
elections and can never be elected to the
infrastructure provisioning
cluster primary. along with auto-scaling, so
teams can get the database
Atlas Search resources they need, when
they need them, and elastically
Atlas Search is built into MongoDB Atlas, scale in response to application
making it easy to build fast, full-text search demands.
capabilities on top of your MongoDB data ° Always-on security to protect
with no need to learn a different API or data, including network
deploy a separate search technology. Atlas isolation, fine-grained access
Search is built on top of Apache Lucene, controls, auditing, and end-to-
the industry standard library. Search end encryption down to the
indexes run alongside the database and level of individual fields.
are automatically kept in sync. Supported ° Certifications with global
search capabilities include fuzzy search, standards for supporting
autocomplete, facets and filters, custom compliance, including ISO
scoring, analyzers for more than 30 27001, SOC 2, and more. Atlas
languages, and more. Database can be used for
workloads subject to HIPAA,
PCI-DSS, or GDPR.
Atlas Data Lake ° Built in replication both within
and across regions for always-
Atlas Data Lake is an on-demand query on availability, even in the face
service that enables you to analyze data of complete regional outages.
in cloud object storage (Amazon S3) in ° Global Clusters for fully
place using the MongoDB Query API. managed, globally distributed
There is no infrastructure to set up or databases that provide low-
manage. Atlas Data Lake automatically
Continued on next page »

13
MongoDB
Architecture Guide

parallelizes operations by breaking down latency, responsive reads

queries and dividing the work across and writes to users anywhere,
multiple compute nodes. Atlas Data with strong data sovereignty
Lake can also automatically optimize controls for regulatory
workloads by utilizing compute in the compliance. Global Clusters
region closest to your data. This is useful allow you to quickly implement
for data residency, granting you the ability zoned sharding using a visual
to specify the region in which your data UI or the Atlas Admin API. Each
should be processed. zone is part of the same cluster,
so they can be queried globally,
Support for federated queries allows you to but data is pinned to shards in
combine and analyze data across S3 and specific regions based on data
your Atlas database clusters, together with localization policies.
a single query. In addition, you can easily ° Multi-cloud clusters allow
persist the results of aggregations to either you to distribute the data in a
object storage or your cloud database. single logical database across
Supported data formats include JSON, multiple cloud providers for
BSON, CSV, TSV, Avro, ORC, and Parquet. cross-cloud redundancy, even
wider geographic reach, and
Atlas Charts seamless migrations across
cloud providers. Multi-cloud
clusters can also be used to
Atlas Charts is a data visualization service
easily leverage the best services
that natively supports richly structured
from each cloud provider on
JSON data. Easily create charts, graphs,
your live, operational data —
and dashboards in a drag-and-drop
e.g., users who run primarily
interface, and share them with other
in AWS can quickly spin up a
users for collaboration or embed them replica on Google Cloud to take
directly into your applications to create advantage of Google’s latest AI/
engaging user experiences. Atlas Charts ML services.
can be configured to read from analytics
° Fully managed backups
or secondary nodes, ensuring no impact
with point-in-time recovery
to operational workloads. Supported
to protect against data
data sources include one or more Atlas corruption, and the ability to
Database deployments, Atlas Data Lake, query backups in place without
or a combination of both. full restores.
° Fine-grained monitoring, real-
Realm Sync time metrics, query profiler,
and customizable alerts for
Realm Sync provides bidirectional data comprehensive performance
sync between Atlas Database clusters and visibility.
Realm, a lightweight, open-source mobile
Continued on next page »

14
MongoDB
Architecture Guide

database. Realm is a more developer- ° Intelligent schema and index

friendly alternative to embedded data recommendations with the
stores such as SQLite or Core Data. Performance Advisor, which
This joint solution helps solve the unique analyzes slow query logs of
challenges of building offline-first your database collections and
applications for mobile, making it simple provides suggestions ranked
to store data on-device — allowing data by impact to your database
access even when offline — and enabling performance.
bidirectional updates when a connection ° Automated patching and
is established. Realm’s SDKs give single-click upgrades for new
developers the tools needed to access major versions of the database,
data stored in MongoDB Atlas directly enabling you to take advantage
from the client and interact with the of the latest MongoDB features.
platform’s broader set of services. ° Auto-archiving of aged data
from your live database clusters
to fully managed cloud object
GraphQL storage with Online Archive.
Federated query enables you
Automatically generate a JSON schema to analyze your data from
for your MongoDB collections and your operational database
enable GraphQL for your MongoDB and historical data on object
apps with a simple click. By querying storage together and in
against a single endpoint to get exactly place with a single query for
the data you need, you can build highly faster insights. Queries are
performant applications. When you use automatically routed to the
GraphQL alongside MongoDB’s other app appropriate data service
development features — such as built-in without having to think about
authentication and data access control — data movement, replication,
it’s also simple to secure your app. or ETL.
° Live migration to move a self-
Event-Driven Architecture managed MongoDB database
into the Atlas service or to
move Atlas databases between
Part of the broader services available
cloud providers.
to Atlas users, functions allow you to
define and execute server-side logic ° A 512 MB perpetual free tier.
without having to provision or manage
Atlas Database is serving a vast range
servers, making it easy to integrate with
of workloads for startups, Fortune
cloud services, build APIs, and more.
500 companies, and government
Atlas Triggers allow you to automatically
agencies, including mission-critical
execute functions in real time — in response
applications handling highly sensitive
to changes in the database or user-
data in regulated industries.
authentication events, or at preset intervals.

15
MongoDB
Architecture Guide

MongoDB for Mission-Critical

Applications in Your
Data Center
If you need to run MongoDB on your own automate deployment, monitoring, backup,
self-managed infrastructure for business and scaling of MongoDB. You can also
or regulatory requirements, MongoDB manage the complete life cycle of your
Enterprise Advanced is a finely tuned MongoDB databases via a powerful GUI,
package of advanced software, support, or programmatically with APIs to enable
certifications, and other services that can integration with your Infrastructure-as-
help. Enterprise Advanced can be used Code (IaC) tools.
to power a MongoDB database behind
Kubernetes users can use the MongoDB
a single application, or to build your own
Enterprise Operator for Kubernetes, which
private database service and expose it to
integrates with MongoDB Ops Manager to
your development teams.
automate and manage MongoDB clusters.
As part of the MongoDB Enterprise It gives you full control over your MongoDB
Advanced subscription, MongoDB deployment from a single Kubernetes
Enterprise Server is a version of the control plane. You can use the operator with
database software that includes an in- upstream Kubernetes, or with any popular
memory storage engine for high throughput distribution such as Red Hat OpenShift or
and predictable low latency; advanced Pivotal Container Service (PKS).
security options, such as LDAP and
The MongoDB Connector for BI lets you
Kerberos access controls; comprehensive
use MongoDB as a data source for your
auditing; and an encrypted storage engine
existing SQL-based BI and analytics
for protecting data at rest.
platforms such as Tableau, Microstrategy,
MongoDB Ops Manager simplifies the Looker, and more. It is included with
administration tasks associated with MongoDB Enterprise Advanced and
running MongoDB on premises or in a available in a pay-as-you-go model for
private cloud. With Ops Manager, you can Atlas database clusters.

16
MongoDB
Architecture Guide

Run MongoDB for Free

With Tools From Us
MongoDB Community Server is the free, It also helps you identify issues before
source-available version of the database they become emergencies and streamline
software. It has been downloaded hundreds operations. With Cloud Manager, you
of millions of times and includes all the core can monitor trends, see live workload
database functionality — flexible document characteristics, set up alerts, and get
model, expressive query API, replication, performance-optimization suggestions.
and sharding — to support building a wide
variety of applications.
Connectors
MongoDB Compass The MongoDB Connector for Apache Spark
exposes all of Spark’s libraries, including
You can easily interact with your MongoDB Scala, Java, Python, and R. MongoDB data
data using MongoDB Compass, the GUI is materialized as DataFrames and Datasets
for MongoDB. Through Compass you for analysis with machine learning, graph,
can explore and manipulate data, create streaming, and SQL APIs.
queries and aggregation pipelines visually
With the MongoDB Connector for Apache
from the GUI and then export them as code
Kafka, you can build robust data pipelines
to your app, view and create indexes, build
that move events between systems in real
schema validation rules, and more.
time, using MongoDB as both a source and
sink for Kafka. The connector is supported
Cloud Manager by MongoDB and verified by Confluent.
You can use any distribution of Kubernetes
Cloud Manager is the cloud-based to manage the full life cycle of your
management platform that enables you MongoDB clusters, wherever you choose to
to deploy, monitor, back up, and scale run them, from on-premises infrastructure
MongoDB. It enables you to automate to the public cloud. With MongoDB’s
administration tasks like deployment, Kubernetes integrations, you can run and
monitoring and alerts, scaling, upgrades, scale your clusters with ease regardless of
backup, and performance optimization. your chosen infrastructure topology.

17
MongoDB
Architecture Guide

Getting Started
Every industry is in the midst of digital In this guide we explored the foundational
transformation. Many businesses are concepts that underpin the architecture of
unable to realize the full potential of MongoDB. Other guides on topics such as
their investments because they fail to performance, operations, and security best
modernize their data architecture. As practices can be found at MongoDB.com.
you build or remake your company for a
digital world, speed matters — measured
by how fast you build applications, scale You can get started now with MongoDB by:
them, and gain insights from the data 1. Reviewing the Use Case Guidance
they generate. These are the keys to White Paper to identify applicable
applications that provide better customer use cases for MongoDB.
experiences; enable deeper, data-
2. Spinning up a fully managed
driven insights; and make new products
MongoDB cluster on the Atlas free
or business models possible. MongoDB
tier or downloading MongoDB for
enables you to meet the demands of local development.
modern apps with a complete application
3. Reviewing the MongoDB manuals and
data platform that includes all the
tutorials in our documentation.
complementary services developers need.

Safe Harbor
The development, release, and timing of any features or functionality described for our
products remains at our sole discretion. This information is merely intended to outline our
general product direction, and it should not be relied on in making a purchasing decision, nor is
this a commitment, promise, or legal obligation to deliver any material, code, or functionality.

© 2021 MongoDB, Inc. MongoDB and the MongoDB leaf logo are
registered trademarks of MongoDB, Inc. Published November 2021.

18