Proprietary + Confidential

Storage and Database

Services

In this module, we cover storage and database services in Google Cloud. Every
application needs to store data, whether it's business data, media to be streamed, or
sensor data from devices.

From an application-centered perspective, the technology stores and retrieves the

data. Whether it's a database or an object store is less important than whether that
service supports the application’s requirements for efficiently storing and retrieving the
data, given its characteristics.

Google offers several data storage services to choose from. In this module, we will
cover Cloud Storage, Filestore, Cloud SQL, Cloud Spanner, AlloyDB, Firestore, Cloud
Bigtable, and Memorystore. Let me start by giving you a high-level overview of these
different services.
 Proprietary + Confidential

Storage and database services

Object File Relational Non-relational Warehouse Redis

Cloud Cloud Cloud

Filestore Cloud SQL AlloyDB Firestore BigQuery Memorystore
Storage Spanner Bigtable
Good for: Good for: Good for: Good for: Good for: Good for: Good for: Good for: Good for:
Binary or Network Web RDBMS + Hybrid Hierarchical, Heavy read Enterprise Automating
object data Attached frameworks scale, HA, transactional mobile, web + write, data complex
Storage HTAP and events, warehouse Redis and
(NAS) analytical Memcached
processing tasks

Such as: Such as: Such as: Such as: Such as: Such as: Such as: Such as: Such as:
Images, Latency CMS, User Machine User AdTech, Analytics, Enabling
media sensitive eCommerce metadata, learning, profiles, financial, dashboards high
serving, workloads Ad/Fin/ Generative AI game state IoT availability,
backups MarTech failover,
patching

This table shows the storage and database services and highlights the storage
service type (object, file, relational, non-relational, data warehouse, and Redis), what
each service is good for, and intended use.

BigQuery is also listed on the right. I’m mentioning this service because it sits on the
edge between data storage and data processing. You can store data in BigQuery, but
the intended use for BigQuery is big data analysis and interactive querying. For this
reason, BigQuery is covered later in the course.
 Proprietary + Confidential

Scope

Infrastructure Track Data Engineering Track

● Service differentiators ● How to use a database system
● When to consider using each service ● Design, organization, structure, schema, and
● Set up and connect to a service use for an application
● Details about how a service stores and
retrieves structured data

Before we dive into each of the data storage services, let’s define the scope of this
module.

The purpose of this module is to explain which services are available and when to
consider using them from an infrastructure perspective. We want you to be able to set
up and connect to a service without detailed knowledge of how to use a database
system.

If you want a deeper dive into the design, organizations, structures, schemas and
details on how data can be optimized, served and stored properly within those
different services, we recommend Google Cloud’s Data Engineering courses.
 Proprietary + Confidential

Agenda

01 Cloud Storage and Filestore

Lab: Cloud Storage

02 Cloud SQL
Lab: Implementing Cloud SQL

03 Cloud Spanner

04 AlloyDB

05 Firestore

06 Cloud Bigtable

07 Memorystore

Let’s look at the agenda. This module covers all of the services we have mentioned
so far. To become more comfortable with these services, you will apply them in two
labs.

I’ll also provide a quick overview of Memorystore, which is Google Cloud’s fully
managed Redis service.

Let’s get started by diving into Cloud Storage and Filestore.

 Proprietary + Confidential

Cloud Storage and

Filestore

01
 Proprietary + Confidential

Cloud Storage is an object storage service

Use cases: Key features:

● Website content ● Scalable to exabytes
● Storing data for archiving and ● Time to first byte in milliseconds
disaster recovery
● Very high availability across all
● Distributing large data objects to storage classes
users via direct download
● Single API across storage classes

Cloud Storage is Google Cloud’s object storage service, and it allows world-wide
storage and retrieval of any amount of data at any time. You can use Cloud Storage
for a range of scenarios including serving website content, storing data for archival
and disaster recovery, or distributing large data objects to users via direct download.
Cloud Storage has a couple of key features:

● It’s scalable to exabytes of data

● The time to first byte is in milliseconds
● It has very high availability across all storage classes
● And It has a single API across those storage classes

Some like to think of Cloud Storage as files in a file system but it’s not really a file
system. Instead, Cloud Storage is a collection of buckets that you place objects into.
You can create directories, so to speak, but really a directory is just another object
that points to different objects in the bucket. You’re not going to easily be able to index
all of these files like you would in a file system. You just have a specific URL to access
objects.
 Proprietary + Confidential

Overview of storage classes

Standard Nearline Coldline Archive

Use case “Hot” data and/or Infrequently Infrequently Data archiving, online
stored for only brief accessed data like accessed data that backup, and disaster
periods of time like data backup, you read or modify at recovery
data-intensive long-tail multimedia most once a quarter
computations content, and data
archiving

Minimum storage None 30 days 90 days 365 days

duration

Retrieval cost None $0.01 per GB $0.02 per GB $0.05 per GB

Availability SLA 99.95% (multi/dual) 99.90% (multi/dual) 99.90% (multi/dual)

99.90% (region) 99.00% (region) 99.00% (region)

Durability 99.999999999%

Cloud Storage has four storage classes: Standard, Nearline, Coldline and Archive and
each of those storage classes provide 3 location types:

● A multi-region is a large geographic area, such as the United States, that

contains two or more geographic places.
● A dual-region is a specific pair of regions, such as Finland and the
Netherlands.
● A region is a specific geographic place, such as London.

Objects stored in a multi-region or dual-region are geo-redundant. Now, let’s go over

each storage class:

● Standard Storage is best for data that is frequently accessed ("hot" data)
and/or stored for only brief periods of time. This is the most expensive storage
class but it has no minimum storage duration and no retrieval cost. When used
in a:
○ region, Standard Storage is appropriate for storing data in the same
location as Google Kubernetes Engine clusters or Compute Engine
instances that use the data. Co-locating your resources maximizes the
performance for data-intensive computations and can reduce network
charges.
 ○ dual-region, you still get optimized performance when accessing
Google Cloud products that are located in one of the associated
regions, but you also get the improved availability that comes from
storing data in geographically separate locations.
○ multi-region, Standard Storage is appropriate for storing data that is
accessed around the world, such as serving website content,
streaming videos, executing interactive workloads, or serving data
supporting mobile and gaming applications.
● Nearline Storage is a low-cost, highly durable storage service for storing
infrequently accessed data like data backup, long-tail multimedia content, and
data archiving. Nearline Storage is a better choice than Standard Storage in
scenarios where slightly lower availability, a 30-day minimum storage duration,
and costs for data access are acceptable trade-offs for lowered at-rest storage
costs.
● Coldline Storage is a very-low-cost, highly durable storage service for storing
infrequently accessed data. Coldline Storage is a better choice than Standard
Storage or Nearline Storage in scenarios where slightly lower availability, a
90-day minimum storage duration, and higher costs for data access are
acceptable trade-offs for lowered at-rest storage costs.
● Archive Storage is the lowest-cost, highly durable storage service for data
archiving, online backup, and disaster recovery. Unlike the "coldest" storage
services offered by other Cloud providers, your data is available within
milliseconds, not hours or days. Archive Storage also has higher costs for data
access and operations, as well as a 365-day minimum storage duration.
Archive Storage is the best choice for data that you plan to access less than
once a year.

Let’s focus on durability and availability. All of these storage classes have 11 nines of
durability, but what does that mean? Does that mean you have access to your files at
all times? No, what that means is you won't lose data. You may not be able to access
the data which is like going to your bank and saying well my money is in there, it's 11
nines durable. But when the bank is closed we don't have access to it which is the
availability that differs between the storage classes and the location type.
 Proprietary + Confidential

Cloud Storage overview

Buckets Object
● Naming requirements
● Cannot be nested

Objects
● Inherit storage class of bucket when created
● No minimum size; unlimited storage

Access
● gcloud storage command
Bucket
● (RESTful) JSON API or XML API

Cloud Storage is broken down into a couple of different items here.

● First of all, there are buckets which are required to have a globally unique
name and cannot be nested.
● The data that you put into those buckets are objects that inherit the storage
class of the bucket and those objects could be text files, doc files, video files,
etc. There is no minimum size to those objects and you can scale this as much
as you want as long as your quota allows it.
● To access the data, you can use the gcloud storage command, or either
the JSON or XML APIs.
 Proprietary + Confidential

Changing default storage classes

● Default class is applied to new objects

● Regional bucket can never be changed to Multi-Region/Dual-Region

● Multi-Regional bucket can never be changed to Regional

● Objects can be moved from bucket to bucket

● Object Lifecycle Management can manage the classes of objects

When you upload an object to a bucket, the object is assigned the bucket's storage
class, unless you specify a storage class for the object. You can change the default
storage class of a bucket but you can't change the location type from regional to
multi-region/dual-region or vice versa.

You can also change the storage class of an object that already exists in your bucket
without moving the object to a different bucket or changing the URL to the object.
Setting a per-object storage class is useful, for example, if you have objects in your
bucket that you want to keep, but that you don't expect to access frequently. In this
case, you can minimize costs by changing the storage class of those specific objects
to Nearline, Coldline or Archive Storage.

In order to help manage the classes of objects in your bucket, Cloud Storage offers
Object Lifecycle Management. More on that later.
 Proprietary + Confidential

Access control

Project
Signed Policy
IAM ACLs Signed URL
Document
Bucket

Signed and Timed

Cryptographic Key Controls File
Identity and Access Upload Policy
Management Access Control Lists

Object

Can be used together

Let’s look at access control for your objects and buckets that are part of a project.

● We can use IAM for the project to control which individual user or service
account can see the bucket, list the objects in the bucket, view the names of
the objects in the bucket, or create new buckets. For most purposes, IAM is
sufficient, and roles are inherited from project to bucket to object.
● Access control lists or ACLs offer finer control.
● For even more detailed control, signed URLs provide a cryptographic key that
gives time-limited access to a bucket or object.
● Finally, a signed policy document further refines the control by determining
what kind of file can be uploaded by someone with a signed URL. Let’s take a
closer look at ACLs and signed URLs.
 Proprietary + Confidential

Access control lists (ACLs)

Scope Permission

ACLs Owner

Max: 100 ACL entries Writer

Examples:
Reader
● [email protected]
● allUsers
● allAuthenticatedUsers

An ACL is a mechanism you can use to define who has access to your buckets and
objects, as well as what level of access they have. The maximum number of ACL
entries you can create for a bucket or object is 100.

Each ACL consists of one or more entries, and these entries consist of two pieces of
information:

● A scope, which defines who can perform the specified actions (for example, a
specific user or group of users).
● And a permission, which defines what actions can be performed (for example,
read or write).

The allUsers identifier listed on this slide represents anyone who is on the internet,
with or without a Google account. The allAuthenticatedUsers identifier, in contrast,
represents anyone who is authenticated with a Google account.

For more information on ACLs, please refer to the documentation.

 Proprietary + Confidential

Signed URLs

● “Valet key” access to buckets and objects via ticket:

○ Ticket is a cryptographically signed URL
○ Time-limited
○ Operations specified in ticket: HTTP GET, PUT, DELETE (not POST)
○ Any user with URL can invoke permitted operations

● Example using private account key and gcloud storage:

gcloud storage signurl -d 10m path/to/privatekey.p12
gs://bucket/object

For some applications, it is easier and more efficient to grant limited-time access
tokens that can be used by any user, instead of using account-based authentication
for controlling resource access. (For example, when you don’t want to require users
to have Google accounts).

Signed URLs allow you to do this for Cloud Storage. You create a URL that grants
read or write access to a specific Cloud Storage resource and specifies when the
access expires. That URL is signed using a private key associated with a service
account. When the request is received, Cloud Storage can verify that the
access-granting URL was issued on behalf of a trusted security principal, in this case
the service account, and delegates its trust of that account to the holder of the URL.

After you give out the signed URL, it is out of your control. So you want the signed
URL to expire after some reasonable amount of time.
 Proprietary + Confidential

Cloud Storage features

● Customer-supplied encryption key (CSEK)

○ Use your own key instead of Google-managed keys
● Object Lifecycle Management
○ Automatically delete or archive objects
● Object Versioning
○ Maintain multiple versions of objects
● Directory synchronization
○ Synchronizes a VM directory with a bucket
● Object change notifications using Pub/Sub
● Autoclass

There are also several features that come with Cloud Storage. We will cover these at
a high-level for now because we will soon dive deeper into some of them.

● Earlier in the course, we already talked a little about Customer-supplied

encryption keys when attaching persistent disks to virtual machines. This
allows you to supply your own encryption keys instead of the
Google-managed keys, which is also available for Cloud Storage.
● Cloud Storage also provides Object Lifecycle Management which lets you
automatically delete or archive objects.
● Another feature is object versioning which allows you to maintain multiple
version of objects in your bucket. You are charged for the versions as if they
were multiple files, which is something to keep in mind.
● Cloud Storage also offers directory synchronization so that you can sync a VM
directory with a bucket.
● Object change notifications can be configured for Cloud Storage using
Pub/Sub.
● When enabled, Autoclass manages all aspects of storage classes for a bucket.
We will discuss this later.
 Proprietary + Confidential

Object Versioning supports the retrieval of objects that

are deleted or overwritten
Cloud Storage Object Versioning
● Objects are immutable.

● Object Versioning:
○ Maintain a history of Bucket Archive

modifications of objects. Object A (g1) Object A (g1)

○ List archived versions of an Object A (g2)

object, restore an object to an New Object A

older state, or delete a version.

In Cloud Storage, objects are immutable, which means that an uploaded object
cannot change throughout its storage lifetime. To support the retrieval of objects that
are deleted or overwritten, Cloud Storage offers the Object Versioning feature. Object
Versioning can be enabled for a bucket. Once enabled, Cloud Storage creates an
archived version of an object each time the live version of the object is overwritten or
deleted. The archived version retains the name of the object but is uniquely identified
by a generation number as illustrated on this slide by g1.

When Object Versioning is enabled, you can list archived versions of an object,
restore the live version of an object to an older state, or permanently delete an
archived version, as needed. You can turn versioning on or off for a bucket at any
time. Turning versioning off leaves existing object versions in place and causes the
bucket to stop accumulating new archived object versions.

For more information on Object Versioning, please refer to the documentation.

 Proprietary + Confidential

Object Lifecycle Management policies specify actions

to be performed on objects that meet certain rules
● Examples:
○ Downgrade storage class on objects older than a year.
○ Delete objects created before a specific date.
○ Keep only the 3 most recent versions of an object.

● Object inspection occurs in asynchronous batches.

● Changes can take 24 hours to apply.

To support common use cases like setting a Time to Live for objects, archiving older
versions of objects, or "downgrading" storage classes of objects to help manage
costs, Cloud Storage offers Object Lifecycle Management.

You can assign a lifecycle management configuration to a bucket. The configuration

is a set of rules that apply to all the objects in the bucket. So when an object meets
the criteria of one of the rules, Cloud Storage automatically performs a specified
action on the object. Here are some example use cases:

● Downgrade the storage class of objects older than a year to Archive Storage.
● Delete objects created before a specific date. For example, January 1, 2017.
● Keep only the 3 most recent versions of each object in a bucket with
versioning enabled.

Object inspection occurs in asynchronous batches, so rules may not be applied

immediately. Also, updates to your lifecycle configuration may take up to 24 hours to
go into effect. This means that when you change your lifecycle configuration, Object
Lifecycle Management may still perform actions based on the old configuration for up
to 24 hours. So keep that in mind.

For more information on Object Lifecycle Management, please refer to the

documentation.
 Proprietary + Confidential

Pub/Sub notifications for Cloud Storage

Cloud Functions

Cloud Storage

Pub/Sub

You can configure object change notifications for Cloud Storage using Pub/Sub.
Pub/Sub notifications sends information about changes to objects in your buckets to
Pub/Sub, where the information is added to a Pub/Sub topic of your choice in the form
of messages. For example, you can track objects that are created and deleted in your
bucket. Each notification contains information describing both the event that triggered
it and the object that changed. You can send notifications to any Pub/Sub topic in any
project for which you have sufficient permissions. Once received by the Pub/Sub
topic, the resulting message can be sent to any number of subscribers to the topic.
Refer to the Prerequisites documentation for information on connecting your Cloud
Storage buckets to a Pub/Sub topic.

Pub/Sub notifications are the recommended way to track changes to objects in your
Cloud Storage buckets because they're faster, more flexible, easier to set up, and
more cost-effective. Pub/Sub is Google’s distributed real-time messaging service,
which is covered in the Developing Applications track.
 Proprietary + Confidential

Data import services

● Transfer Appliance: Rack, capture and then

ship your data to Google Cloud.

● Storage Transfer Service: Import online data

(another bucket, an S3 bucket, or web source).

● Offline Media Import: Third-party provider

uploads the data from physical media.

The Google Cloud console allows you to upload individual files to your bucket. What if
you have to upload terabytes or even petabytes of data? There are three services that
address this: Transfer Appliance, Storage Transfer Service, and Offline Media Import.

● Transfer Appliance is a hardware appliance you can use to securely migrate

large volumes of data (from hundreds of terabytes up to 1 petabyte) to Google
Cloud without disrupting business operations. The images on this slide are
transfer appliances.
● The Storage Transfer Service enables high-performance imports of online
data. That data source can be another Cloud Storage bucket, an Amazon S3
bucket, or an HTTP/HTTPS location.
● Finally, Offline Media Import is a third party service where physical media
(such as storage arrays, hard disk drives (HDDs), tapes, and USB flash
drives) is sent to a provider who uploads the data.
 Proprietary + Confidential

Cloud Storage provides strong global consistency

● Read-after-write

● Read-after-metadata-update

● Read-after-delete

● Bucket listing

● Object listing

When you upload an object to Cloud Storage and you receive a success response,
the object is immediately available for download and metadata operations from any
location where Google offers service. This is true whether you create a new object or
overwrite an existing object. Because uploads are strongly consistent, you will never
receive a 404 Not Found response or stale data for a read-after-write or
read-after-metadata-update operation.

Strong global consistency also extends to deletion operations on objects. If a deletion

request succeeds, an immediate attempt to download the object or its metadata will
result in a 404 Not Found status code. You get the 404 error because the object no
longer exists after the delete operation succeeds.

Bucket listing is strongly consistent. For example, if you create a bucket, then
immediately perform a list buckets operation, the new bucket appears in the returned
list of buckets.
Finally, object listing is also strongly consistent. For example, if you upload an object
to a bucket and then immediately perform a list objects operation, the new object
appears in the returned list of objects.
 Proprietary + Confidential

Choosing a storage class

Start

Structured or unstructured Read < 1 Read < 1 per 90 Read < 1 per 30
unstructured No
per year? days? days?
data?
structured
Yes Yes Yes

Consider a
Consider Consider Consider Standard
structured
Archive Storage Coldline Storage Nearline Storage Storage
database service

Choose location and type by balancing latency, availability, and

bandwidth costs for data consumers

Let’s explore the decision tree to help you find the appropriate storage class in Cloud
Storage.

● If you will read your data less than once a year, you should consider using
Archive storage.
● If you will read your data less than once per 90 days, you should consider
using Coldline storage.
● If you will read your data less than once per 30 days, you should consider
using Nearline storage.
● And if you will be doing reads and writes more often than that, you should
consider using Standard storage.

You also want to take into account the location type:

● Use a region to help optimize latency and network bandwidth for data
consumers, such as analytics pipelines, that are grouped in the same region.
● Use a dual-region when you want similar performance advantages as
regions, but also want the higher availability that comes with being
geo-redundant.
● Use a multi-region when you want to serve content to data consumers that
are outside of the Google network and distributed across large geographic
 ● areas, or when you want the higher availability that comes with being
geo-redundant.

If your data has a variety of access frequencies, or the access patterns for your data
are unknown or unpredictable, you should consider Autoclass.
 Proprietary + Confidential

Autoclass storage in Google Cloud

Object is read

Autoclass transitions
Standard Nearline Coldline Archive
objects in your bucket to
appropriate storage classes
based on the access
Objects not Objects not Objects not
pattern of each object. accessed for 30 accessed for 90 accessed for Objects stay
days move to days move to 365 days move until accessed
Nearline Coldline to Archive

Objects stay Objects stay Objects stay Objects stay

for 30 days for 60 days for 275 days until accessed

The Autoclass feature automatically transitions objects in your bucket to appropriate

storage classes based on the access pattern of each object. Even if a different
storage class is specified in the request, all objects added to the bucket begin in
Standard storage. The feature moves data that is not accessed to colder storage
classes to reduce storage cost. Data that is accessed is also moved to Standard
storage to optimize future accesses.

When object data is read, the object transitions to Standard storage if it's not already
stored in Standard storage. Autoclass simplifies and automates cost saving for your
Cloud Storage data. When enabled on a bucket, there are no early deletion charges,
no retrieval charges, and no charges for storage class transitions.

For more information, see storage classes.

So far we have only considered unstructured data. Before we look at unstructured

data, let's explore a high-performance, fully managed file storage offering; Filestore.
 Proprietary + Confidential

Filestore is a managed file storage service

for applications
Filestore
● Fully managed network attached storage (NAS) for
Compute Engine and GKE instances.

● Predictable performance.

● Full NFSv3 support.

● Scales to 100s of TBs for high-performance workloads.

Filestore is a managed file storage service for applications that require a filesystem
interface and a shared filesystem for data. Filestore gives users a simple, native
experience for standing up managed Network Attached Storage (NAS) with their
Compute Engine and Google Kubernetes Engine instances. The ability to fine-tune
Filestore’s performance and capacity independently leads to predictably fast
performance for your file-based workloads.

Filestore offers native compatibility with existing enterprise applications and supports
any NFSv3-compatible client. Applications gain the benefit of features such as
scaleout performance, 100s of TBs of capacity, and file locking, without the need to
install or maintain any specialized plugins or client side software.
 Proprietary + Confidential

Filestore has many use cases

Filestore
● Application migration

● Media rendering

● Electronic Design Automation (EDA)

● Data analytics

● Genomics processing

● Web content management

Filestore has many use cases. Using Filestore, you can expedite migration of
enterprise applications. Many on-premises applications require a filesystem interface
to data. As these applications continue to migrate to the cloud, Filestore can support a
broad range of enterprise applications that need a shared filesystem.

For media rendering, you can easily mount Filestore file shares on Compute Engine
instances, enabling visual effects artists to collaborate on the same file share. As
rendering workflows typically run across fleets (“render farms”) of compute machines,
all of which mount a shared filesystem, Filestore and Compute Engine can scale to
meet your job’s rendering needs.

Electronic Design Automation (EDA) is all about data management. It requires the
ability to batch workloads across thousands of cores and has large memory needs.
Filestore offers the necessary capacity and scale to meet the needs of manufacturing
customers doing intensive EDA and also makes sure files are universally accessible.

Data analytics workloads include compute complex financial models or analysis of

environmental data. These workloads are latency sensitive. Filestore offers low
latency for file operations and, as capacity or performance needs change, you can
easily grow or shrink your instances as needed. As a persistent and shareable
storage layer, Filestore enables immediate access to data for high-performance,
smart analytics without the need to lose valuable time on loading and off-loading data
to clients’ drives.

Genome sequencing requires an incredible amount of raw data, on the order of

billions of data points per person. This type of analysis requires speed, scalability, and
security. Filestore meets the needs of companies and research institutions performing
scientific research, while also offering predictable prices for the performance.

Web developers and large hosting providers rely on Filestore to manage and serve
web content, including needs such as WordPress hosting.
 Proprietary + Confidential

Lab Intro
Cloud Storage

Let’s take some of the Cloud Storage concepts that we just discussed and apply them
in a lab.
 Proprietary + Confidential

Lab objectives

01 Create and use buckets

02 Set access control lists to restrict access

03 Use your own encryption keys

04 Implement version controls

05 Use directory synchronization

06 Share a bucket across projects using IAM

In this lab, you'll create buckets and perform many of the advanced options available
in Cloud Storage. You'll set access control lists to limit who can have access to your
data and what they're allowed to do with it. You'll use the ability to supply and manage
your own encryption keys for additional security. You'll enable object versioning to
track changes in the data, and you'll configure lifecycle management so that objects
are automatically archived or deleted after a specified period. Finally, you’ll use the
directory synchronization feature that we mentioned and share your buckets across
projects using IAM.
 Proprietary + Confidential

02
Cloud SQL

Let’s dive into the structured or relational database services. First up is Cloud SQL.
 Proprietary + Confidential

Build your own database solution or use

a managed service

Storage
DB

Compute Cloud Cloud Cloud Cloud Firestore

Engine Bigtable Storage SQL Spanner

Why would you use a Google Cloud service for SQL, when you can install a SQL
Server application image on a VM using Compute Engine?

The question really is, should you build your own database solution or use a managed
service? There are benefits to using a managed service, so let’s learn about why
you’d use Cloud SQL as a managed service inside of Google Cloud.
 Proprietary + Confidential

Cloud SQL is a fully managed database service

(MySQL, PostgreSQL, or Microsoft SQL Server)
Cloud SQL
● Patches and updates automatically applied

● You administer MySQL users

● Cloud SQL supports many clients

○ gcloud sql
○ App Engine, Google Workspace scripts
○ Applications and tools
■ SQL Workbench, Toad
■ External applications using standard MySQL drivers

Cloud SQL is a fully managed service of either MySQL, PostgreSQL, or Microsoft

SQL Server databases. This means that patches and updates are automatically
applied but you still have to administer MySQL users with the native authentication
tools that come with these databases.

Cloud SQL supports many clients, such as Cloud Shell, App Engine and Google
Workspace scripts. It also supports other applications and tools that you might be
used to like SQL Workbench, Toad and other external applications using standard
MySQL drivers.
 Proprietary + Confidential

Cloud SQL instance

Performance:
● 64 TB of storage
● 60,000 IOPS
● 624 GB of RAM
● Scale out with read replicas

Choice:
● MySQL 5.6, 5.7, or 8.0 (default)
● PostgreSQL 9.6, 10, 11, 12, 13, 14 or 15 (default)
● Microsoft SQL Server 2017 or 2019 (Standard default)

Cloud SQL delivers high performance and scalability with up to 64 TB of storage

capacity, 60,000 IOPS, and 624 GB of RAM per instance. You can easily scale up to
96 processor cores and scale out with read replicas.

Currently, you can use Cloud SQL with either MySQL 5.6, 5.7, or 8.0, PostgreSQL
9.6, 10, 11, 12, 13, 14 or 15, or either of the Web, Express, Standard or Enterprise
SQL Server 2017 or 2019 editions.
 Proprietary + Confidential

Cloud SQL services

Region 1

● HA configuration Zone A
Servers Disks

● Backup service Primary instance

Persistent disk 01

Cloud SQL

● Import/export
Regional
● Scaling Client
application
IP address
Persistent disk
○ Up: Machine capacity
○ Out: Read replicas Zone B
Servers Disks

Standby instance
Persistent disk 02

Cloud SQL

Let’s focus on some other services provided by Cloud SQL:

● In HA configuration, within a regional instance, the configuration is made up of

a primary instance and a standby instance. Through synchronous replication
to each zone's persistent disk, all writes made to the primary instance are
replicated to disks in both zones before a transaction is reported as
committed. In the event of an instance or zone failure, the persistent disk is
attached to the standby instance, and it becomes the new primary instance.
Users are then rerouted to the new primary. This process is called a failover.
● Cloud SQL also provides automated and on-demand backups with
point-in-time recovery.
● You can import and export databases using mysqldump, or import and export
CSV files.
● Cloud SQL can also scale up, which does require a machine restart or scale
out using read replicas. That being said, if you are concerned about horizontal
scalability, you’ll want to consider Cloud Spanner which we’ll cover later in this
module.
 Proprietary + Confidential

Connecting to a Cloud SQL instance

Cloud SQL
Connection

Connecting from Connecting from

outside Google within Google
Cloud Cloud

Yes Yes

Yes Need manual

control over SSL
certificates

No, would like

Manual SSL automation Cloud SQL Cloud SQL
Connection Auth Proxy Private IP

Cannot use
SSL Authorized
Networks

Choosing a connection type to your Cloud SQL instance will influence how secure,
performant, and automated it will be. If your are connecting an application that is
hosted within the same Google Cloud project as your Cloud SQL instance, and it is
collocated in the same region, choosing the Private IP connection will provide you
with the most performant and secure connection using private connectivity. In other
words, traffic is never exposed to the public internet. Note: Connecting to the Cloud
SQL Private IP address from VMs in the same region is only a performance-based
recommendation and not a requirement.

If the application is hosted in another region or project, or if you are trying to connect
to your Cloud SQL instance from outside of Google Cloud, you have 3 options. In this
case, we recommend using Cloud SQL Auth Proxy, which handles authentication,
encryption, and key rotation for you. If you need manual control over the SSL
connection, you can generate and periodically rotate the certificates yourself.
Otherwise, you can use an unencrypted connection by authorizing a specific IP
address to connect to your SQL server over its external IP address.

You will explore these options in the upcoming lab, but if you want to learn more about
Private IP, please refer to the documentation.
 Proprietary + Confidential

Choosing Cloud SQL

Do you need
Is your data
in-memory Start
relational?
store?
Do you need a
document
database?
Is your
workload
analytics?
Memorystore
Cloud Bigtable Firestore

Do you need
99.999% BigQuery
Cloud Spanner availability?

NO YES
Cloud SQL

To summarize, let’s explore this decision tree to help you find the right data storage
service with full relational capability.

Memorystore provides a fully-managed in-memory data store service for workloads

requiring microsecond response times, or that have large spikes in traffic, as seen in
gaming environments and real-time analytics.

If you don’t need an in-memory data store, but your use case is relational data used
primarily for analytics, these workloads are best supported by BigQuery.

However, if your relational data workload isn’t analytics and a database with ACID
properties is a requirement, the choice lies between Cloud Spanner and Cloud SQL.

If you don’t need horizontal scaling or a globally available system, Cloud SQL is a
cost-effective solution.

If Cloud SQL as a managed service is better than using or re-implementing your

existing MySQL solution, refer to this solution on how to migrate from MySQL to Cloud
SQL.
 Proprietary + Confidential

Lab Intro
Implementing Cloud SQL

Let’s take some of the Cloud SQL concepts that we just discussed and apply them in
a lab.
 Proprietary + Confidential

Lab objectives

01 Create a Cloud SQL database

02 Configure a virtual machine to run a proxy

Create a connection between an

03 application and Cloud SQL

Connect an application to Cloud SQL

04 using Private IP address

In this lab, you configure a Cloud SQL server and learn how to connect an application
to it via a proxy over an external connection. You also configure a connection over a
Private IP link that offers performance and security benefits. The app we chose to
demonstrate in this lab is Wordpress, but the information and best practices are
applicable to any application that needs a SQL Server.
 Proprietary + Confidential

VPC
europe-west1 us-central1

Internal
Wordpress-europe- Wordpress-us-
proxyIPinstance private-ip

Proxy Private IP
Internal IP
127.0.0.1 Cloud SQL

External IP External IP
Address Address

Encrypted connection

By the end of this lab, you will have 2 working instances of a Wordpress frontend
connected over 2 different connection types to its SQL instance backend, as shown in
this diagram.
 Proprietary + Confidential

03 Cloud Spanner

If Cloud SQL does not fit your requirements because you need horizontal scalability,
consider using Cloud Spanner.
 Proprietary + Confidential

Cloud Spanner combines the benefits of

relational database structure with
non-relational horizontal scale Cloud Spanner

● Scale to petabytes

● Strong consistency

● High availability

● Used for financial and inventory applications

● Monthly uptime
○ Multi-regional: 99.999%
○ Regional: 99.99%

Cloud Spanner is a service built for the cloud specifically to combine the benefits of
relational database structure with non-relational horizontal scale.

This service can provide petabytes of capacity and offers transactional consistency at
global scale, schemas, SQL, and automatic, synchronous replication for high
availability. Use cases include financial applications and inventory applications
traditionally served by relational database technology.

Depending on whether you create a multi-regional or regional instance, you’ll have

different monthly uptime SLAs as shown on this slide. However, for up-to-date
numbers, you should always refer to the documentation.
 Proprietary + Confidential

Characteristics

Cloud Spanner Relational DB Non-Relational DB

Schema Yes Yes No

SQL Yes Yes No

Consistency Strong Strong Eventual

Availability High Failover High

Scalability Horizontal Vertical Horizontal

Replication Automatic Configurable Configurable

Let’s compare Cloud Spanner with both relational and non-relational databases. Like
a relational database, Cloud Spanner has schema, SQL, and strong consistency.
Also, like a non-relational database, Cloud Spanner offers high availability, horizontal
scalability, and configurable replication.

As mentioned, Cloud Spanner offers the best of the relational and non-relational
worlds. These features allow for mission-critical uses cases, such as building
consistent systems for transactions and inventory management in the financial
services and retail industries. To better understand how all of this works, let’s look at
the architecture of Cloud Spanner.
 Proprietary + Confidential

Cloud Spanner architecture

Cloud Spanner instance

Zone 1 Zone 2 Zone 3

DB 1 DB 1 DB 1

DB 2 DB 2 DB 2

A Cloud Spanner instance replicates data in N cloud zones, which can be within one
region or across several regions. The database placement is configurable, meaning
you can choose which region to put your database in. This architecture allows for high
availability and global placement.
 Proprietary + Confidential

Data replication is synchronized across zones using

Google’s global fiber network
Update

Zone 1 Zone 2 Zone 3

Table 1 Table 1 Table 1

Table 2 Table 2 Table 2

The replication of data will be synchronized across zones using Google’s global fiber
network. Using atomic clocks ensures global consistency whenever you are updating
your data.

That’s as far as we’re going to go with Cloud Spanner. Because the focus of this
module is to understand the circumstances when you would use Cloud Spanner, let’s
look at a decision tree.
 Proprietary + Confidential

Choosing Cloud Spanner

start

Need full
No No No No No relational
capability?

No Yes
Outgrown Are you sharding Need Global data +
single instance for DB transactional strong DB
RDBMS? throughput? consistency? consistency? consolidation?*
Consider a Consider
NoSQL service Cloud SQL

Yes Yes Yes Yes Yes

Consider Cloud Spanner

If you have outgrown any relational database, are sharding your databases for
throughput high performance, need transactional consistency, global data and strong
consistency, or just want to consolidate your database, consider using Cloud
Spanner.

If you don’t need any of these, nor full relational capabilities, consider a NoSQL
service such as Firestore, which we will cover next.

If you’re now convinced that using Cloud Spanner as a managed service is better
than using or re-implementing your existing MySQL solution, please refer to this
solution on how to migrate from MySQL to Cloud Spanner.
 Proprietary + Confidential

AlloyDB

04
Let’s now talk about AlloyDB.
 Proprietary + Confidential

AlloyDB is a fully managed

database service
AlloyDB
● Fully managed database service

● Fast transactional processing

● High availability

● Real-time business insights

AlloyDB for PostgreSQL is a fully managed, PostgreSQL-compatible database service

that's designed for your most demanding workloads, including hybrid transactional
and analytical processing. AlloyDB pairs a Google-built database engine with a
cloud-based, multi-node architecture to deliver enterprise-grade performance,
reliability, and availability.

AlloyDB automates administrative tasks, such as backups, replication, patching, and

capacity management. AlloyDB also uses adaptive algorithms and machine learning
for PostgreSQL vacuum management, storage and memory management, data
tiering, and analytics acceleration.

AlloyDB provides fast transactional processing, more than 4X faster than standard
PostgreSQL for transactional workloads. It's suitable for the most demanding
enterprise workloads, including workloads that require high transaction throughput,
large data sizes, or multiple read replicas.

AlloyDB offers provides high-availability and an 99.99% uptime SLA, inclusive of

maintenance. AlloyDB offers provides real-time business insights and is up to 100X
faster than standard PostgreSQL for analytical queries. Built-in integration with Vertex
AI, Google's artificial intelligence platform, lets you call machine learning models.
 Proprietary + Confidential

Firestore

05
If you are looking for a highly-scalable NoSQL database for your applications,
consider using Firestore.
 Proprietary + Confidential

Firestore is a NoSQL document database

Firestore
● Simplifies storing, syncing, and querying data

● Mobile, web, and IoT apps at global scale

● Live synchronization and offline support

● Security features

● ACID transactions

● Multi-region replication

● Powerful query engine

Firestore is a fast, fully managed, serverless, cloud-native NoSQL document

database that simplifies storing, syncing, and querying data for your mobile, web, and
IoT apps at global scale. Its client libraries provide live synchronization and offline
support, and its security features and integrations with Firebase and Google Cloud
accelerate building truly serverless apps.

Firestore also supports ACID transactions, so if any of the operations in the

transaction fail and cannot be retried, the whole transaction will fail.

Also, with automatic multi-region replication and strong consistency, your data is safe
and available, even when disasters strike. Firestore even allows you to run
sophisticated queries against your NoSQL data without any degradation in
performance. This gives you more flexibility in the way you structure your data.
 Proprietary + Confidential

Firestore is the next generation of Datastore

Datastore mode (new server projects):

● Compatible with Datastore applications
● Strong consistency
● No entity group limits

Native mode (new mobile and web apps):

● Strongly consistent storage layer
● Collection and document data model
● Real-time updates
● Mobile and Web client libraries

Firestore is actually the next generation of Datastore. Firestore can operate in

Datastore mode, making it backwards- compatible with Datastore. By creating a
Firestore database in Datastore mode, you can access Firestore's improved storage
layer while keeping Datastore system behavior.

This removes the following Datastore limitations:

● Queries are no longer eventually consistent; instead, they are all strongly
consistent.
● Transactions are no longer limited to 25 entity groups.
● Writes to an entity group are no longer limited to 1 per second.

Firestore in Native mode introduces new features such as:

● A new, strongly consistent storage layer

● A collection and document data model
● Real-time updates
● Mobile and Web client libraries

Firestore is backward compatible with Datastore, but the new data model, real-time
updates, and mobile and web client library features are not. To access all of the new
Firestore features, you must use Firestore in Native mode. A general guideline is to
use Firestore in Datastore mode for new server projects, and Native mode for new
mobile and web apps.
As the next generation of Datastore, Firestore is compatible with all Datastore APIs
and client libraries. Existing Datastore users will be live-upgraded to Firestore
automatically at a future date. For more information, please refer to these links:

https://cloud.google.com/datastore/docs/firestore-or-datastore
https://cloud.google.com/datastore/docs/upgrade-to-firestore
 Proprietary + Confidential

Choosing Firestore

Start

Schema might change Want low maintenance

Scale down No Transactional consistency
and need an adaptable overhead scaling
to zero? required?
database? up to TBs?

cost / size
Yes No

Consider Firestore Consider Cloud Bigtable

To summarize, let’s explore this decision tree to help you determine whether Firestore
is the right storage service for your data.

If your schema might change and you need an adaptable database, you need to scale
to zero, or you want low maintenance overhead scaling up to terabytes, consider
using Firestore.

Also, if you don’t require transactional consistency, you might want to consider Cloud
Bigtable, depending on the cost or size.

We will cover Cloud Bigtable next.

 Proprietary + Confidential

06
Cloud Bigtable

If you don’t require transactional consistency, you might want to consider Cloud
Bigtable.
 Proprietary + Confidential

Cloud Bigtable is a NoSQL big data

database service
Cloud Bigtable
● Petabyte-scale

● Consistent sub-10ms latency

● Seamless scalability for throughput

● Learns and adjusts to access patterns

● Ideal for Ad Tech, Fintech, and IoT

● Storage engine for ML applications

● Easy integration with open source big data tools

Cloud Bigtable is a fully managed NoSQL database with petabyte-scale and very low
latency. It seamlessly scales for throughput and it learns to adjust to specific access
patterns. Cloud Bigtable is actually the same database that powers many of Google’s
core services, including Search, Analytics, Maps, and Gmail.

Cloud Bigtable is a great choice for both operational and analytical applications,
including IoT, user analytics, and financial data analysis, because it supports high
read and write throughput at low latency. It’s also a great storage engine for machine
learning applications.

Cloud Bigtable integrates easily with popular big data tools like Hadoop, Dataflow,
and Dataproc. Plus, Cloud Bigtable supports the open source industry standard
HBase API, which makes it easy for your development teams to get started. Dataflow
and Dataproc are covered late in the course. For more information on the HBase API,
please refer to this link.
 Proprietary + Confidential

Cloud Bigtable storage model

"follows" column family

Follows
Row Key gwashington jadams tjefferson wmckinley
gwashington 1
jadams 1 1
tjefferson 1 1 1
wmckinley 1

multiple versions

Cloud Bigtable stores data in massively scalable tables, each of which is a sorted
key/value map. The table is composed of rows, each of which typically describes a
single entity, and columns, which contain individual values for each row. Each row is
indexed by a single row key, and columns that are related to one another are typically
grouped together into a column family. Each column is identified by a combination of
the column family and a column qualifier, which is a unique name within the column
family.

Each row/column intersection can contain multiple cells, or versions, at different

timestamps, providing a record of how the stored data has been altered over time.
Cloud Bigtable tables are sparse; if a cell does not contain any data, it does not take
up any space.

The example shown here is for a hypothetical social network for United States
presidents, where each president can follow posts from other presidents. Let me
highlight some things:

● The table contains one column family, the follows family. This family contains
multiple column qualifiers.
● Column qualifiers are used as data. This design choice takes advantage of the
sparseness of Cloud Bigtable tables, and the fact that new column qualifiers
● can be added as your data changes..
● The username is used as the row key. Assuming usernames are evenly
spread across the alphabet, data access will be reasonably uniform across the
entire table.
 Proprietary + Confidential

Processing is separated from storage

Clients

Bigtable Bigtable Bigtable

Processing
node node node

Storage

Colossus file system

This diagram shows a simplified version of Cloud Bigtable’s overall architecture. It

illustrates that processing, which is done through a front-end server pool and nodes,
is handled separately from the storage.

A Cloud Bigtable table is sharded into blocks of contiguous rows, called tablets, to
help balance the workload of queries. Tablets are similar to HBase regions, for those
of you who have used the HBase API.

Tablets are stored on Colossus, which is Google's file system, in SSTable format. An
SSTable provides a persistent, ordered immutable map from keys to values, where
both keys and values are arbitrary byte strings.
 Proprietary + Confidential

Learns access patterns

Clients

Bigtable Bigtable Bigtable

Processing
node node node

Storage A B C D
E
Colossus file system

As we mentioned earlier, Cloud Bigtable learns to adjust to specific access patterns. If

a certain Bigtable node is frequently accessing a certain subset of data...
 Proprietary + Confidential

Rebalances without moving data

Clients

Bigtable Bigtable Bigtable

Processing
node node node

Storage A B C D
E
Colossus file system

… Cloud Bigtable will update the indexes so that other nodes can distribute that
workload evenly, as shown here.
 Proprietary + Confidential

Throughput scales linearly

QPS QPS QPS

80,000 400,000 4m

60,000 300,000 3m

40,000 200,000 2m

20,000 100,000 1m

0 0 0
0 2 4 6 8 0 10 20 30 40 0 100 200 300 400
Bigtable Nodes Bigtable Nodes Bigtable Nodes

That throughput scales linearly, so for every single node that you do add, you're going
to see a linear scale of throughput performance, up to hundreds of nodes.
 Proprietary + Confidential

Choosing Cloud Bigtable

start

Very high r/w latency < 10

Storing > 1 TB
No volume of millisecond and strong HBase API compatible?
structured data?
writes? consistency?

Yes
No

Consider Cloud
Consider Firestore
Bigtable

● Bigtable scales UP well

● Firestore scales DOWN well

In summary, if you need to store more than 1 TB of structured data, have very high
volume of writes, need read/write latency of less than 10 milliseconds along with
strong consistency, or need a storage service that is compatible with the HBase API,
consider using Cloud Bigtable.

If you don’t need any of these and are looking for a storage service that scales down
well, consider using Firestore.

Speaking of scaling, the smallest Cloud Bigtable cluster you can create has three
nodes and can handle 30,000 operations per second. Remember that you pay for
those nodes while they are operational, whether your application is using them or not.
 Proprietary + Confidential

07 Memorystore

Let me give you a quick overview of Memorystore.

 Proprietary + Confidential

Memorystore is a fully managed

Redis service
Memorystore
● In-memory data store service

● Focus on building great apps

● High availability, failover, patching, and monitoring

● Sub-millisecond latency

● Instances up to 300 GB

● Network throughput of 12 Gbps

● Easy Lift-and-Shift

Memorystore for Redis provides a fully managed in-memory data store service built
on scalable, secure, and highly available infrastructure managed by Google.
Applications running on Google Cloud can achieve extreme performance by
leveraging the highly scalable, available, secure Redis service without the burden of
managing complex Redis deployments. This allows you to spend more time writing
code so that you can focus on building great apps.

Memorystore also automates complex tasks like enabling high availability, failover,
patching, and monitoring. High availability instances are replicated across two zones
and provide a 99.9% availability SLA.

You can easily achieve the sub-millisecond latency and throughput your applications
need. Start with the lowest tier and smallest size, and then grow your instance
effortlessly with minimal impact to application availability. Memorystore can support
instances up to 300 GB and network throughput of 12 Gbps.

Because Memorystore for Redis is fully compatible with the Redis protocol, you can
lift and shift your applications from open source Redis to Memorystore without any
code changes by using the import/export feature. There is no need to learn new tools
because all existing tools and client libraries just work.
 Proprietary + Confidential

Quiz
 Proprietary + Confidential

Question #1
Question

What data storage service might you select if you just needed to migrate a standard
relational database running on a single machine in a data center to the cloud?
A. Cloud SQL
B. BigQuery
C. Persistent Disk
D. Cloud Storage
 Proprietary + Confidential

Question #1
Answer

What data storage service might you select if you just needed to migrate a standard
relational database running on a single machine in a data center to the cloud?
A. Cloud SQL
B. BigQuery
C. Persistent Disk
D. Cloud Storage

Explanation:
Cloud SQL offers a PostgreSQL server or a MySQL server as a managed service.
 Proprietary + Confidential

Question #2
Question

Which Google Cloud data storage service offers ACID transactions and can scale globally?
A. Cloud Storage
B. Cloud CDN
C. Cloud Spanner
D. Cloud SQL
 Proprietary + Confidential

Question #2
Answer

Which Google Cloud data storage service offers ACID transactions and can scale globally?
A. Cloud Storage
B. Cloud CDN
C. Cloud Spanner
D. Cloud SQL

Explanation:
Cloud Spanner provides ACID (Atomicity, Consistency, Isolation, Durability) properties
that enable transactional reads and writes on the database. It can also scale globally.
 Proprietary + Confidential

Question #3
Question

Which data storage service provides data warehouse service for storing data but also
offers an interactive SQL interface for querying the data?
A. BigQuery
B. Dataproc
C. Datalab
D. Cloud SQL
 Proprietary + Confidential

Question #3
Answer

Which data storage service provides data warehouse service for storing data but also
offers an interactive SQL interface for querying the data?
A. BigQuery
B. Dataproc
C. Datalab
D. Cloud SQL

Explanation:
BigQuery is a data warehousing service that allows the storage of huge data sets
while making them immediately processable without having to extract or run the
processing in a separate service.
 Proprietary + Confidential

Review: Storage and

Database Services
 Proprietary + Confidential

Object File Relational Non-relational Redis

Cloud Cloud Cloud Memorystore

Filestore Cloud SQL AlloyDB Firestore
Storage Spanner Bigtable

In this module, we covered the different storage and database services that Google
Cloud offers. Specifically, you learned about Cloud Storage, a fully managed object
store; Filestore, a fully managed file storage service; Cloud SQL, a fully managed
MySQL and PostgreSQL database service; Cloud Spanner, a relational database
service with transactional consistency, global scale and high availability; AlloyDB, a
fully managed, PostgreSQL-compatible database service; Firestore, a fully managed
NoSQL document database; Cloud Bigtable, a fully managed NoSQL wide-column
database; and Memorystore, a fully managed in-memory data store service for Redis.

From an infrastructure perspective, the goal was to understand what services are
available and how they're used in different circumstances. Defining a complete data
strategy is beyond the scope of this course; however, Google offers courses on data
engineering and machine learning on Google Cloud that cover data strategy. For
more information, refer to the Google Cloud training site.
 Proprietary + Confidential

Decision chart
Is your data
structured?
Start NO YES

Do you need a
shared file Does your
system? workload involve
analytics?

Filestore Cloud Storage Do you need

extensive updates
Is your data and/or low latency?
relational?

Do you need
HTAP?
Do you need Bigtable BigQuery
application
caching? NoSQL with high Data
throughput Warehouse with
Do you SQL querying
AlloyDB need global
scalability?
Memorystore Firestore

Cloud Cloud
Spanner SQL

Let’s summarize the services in this module with this decision chart:

● First, ask yourself: Is your data structured? If its not, then ask yourself if you
need a shared file system. If you do, then choose Filestore.
● If you don't, then choose Cloud Storage.
● If your data is structured, does your workload focus on analytics? If it does,
you will want to choose Bigtable or BigQuery, depending on your latency and
update needs.
○ BigQuery is recommended as a data warehouse, is the default storage
for tabular data, and is optimized for large-scale, ad-hoc SQL-based
analysis and reporting. While BigQuery data manipulation language
(DML) enables you to update, insert, and delete data from your
BigQuery tables, because it has a built-in cache BigQuery works really
well in cases where the data does not change often.
○ Bigtable is a NoSQL wide-column database. It's optimized for low
latency, large numbers of reads and writes, and maintaining
performance at scale.
○ In addition to analytics, Bigtable is also suited as a ‘fast lookup’
non-relational database for datasets too large to store in memory, with
use cases in areas such as IoT, AdTech and FinTec.
● If your workload doesn’t involve analytics, check whether your data is
relational. If it’s not relational, do you need application caching?
○ If caching is a requirement, choose Memorystore, an in-memory
database.
 ○ Otherwise choose Firestore, a document database.
● If your data is relational and you need HTAP(Hybrid transaction/analytical
processing) choose AlloyDB.
○ If you don’t need HTAP (Hybrid transaction/analytical processing) and
don’t need global scalability, choose Cloud SQL.
○ If you don’t need HTAP (Hybrid transaction/analytical processing) and
need global scalability, choose Spanner.

Depending on your application, you might use one or several of these services to get
the job done. For more information on how to choose between these different
services, please refer to the following two links:

https://cloud.google.com/storage-options/
https://cloud.google.com/products/databases/