This page describes how to migrate from Amazon Simple Storage Service (Amazon S3) to Google Cloud Storage for users sending requests using an API. If you are not currently using Amazon S3 and you want to send requests using the Cloud Storage API, then start here: XML API Overview.
If you are new to Google Cloud Storage and will not be using the API directly, consider using the Google Cloud Platform Console. The Google Cloud Platform Console provides a graphical interface to Cloud Storage that enables you to accomplish many of your storage tasks using just a browser.
Overview of migration
If you are an Amazon S3 user, you can easily migrate your applications that use Amazon S3 to use Google Cloud Storage. You have two migration options:
- Simple Migration
This is this easiest way to get started with Cloud Storage if you are coming from Amazon S3 because it requires just a few simple changes to the tools and libraries you currently use with Amazon S3. For more information, see Simple Migration.
While a simple migration allows you to get going quickly with Cloud Storage, it does not allow you to use all the features of Google Cloud Storage. To take full advantage of Google Cloud Storage, follow the steps for a full migration.
- Full Migration
A full migration from Amazon S3 to Google Cloud Storage requires a few extra steps than a simple migration, but the benefit is that you can use all the features of Google Cloud Storage, including support for service accounts, multiple projects, and OAuth 2.0 for authentication. For more information, see Full Migration.
In a simple migration from Amazon S3 to Google Cloud Storage, you can use your existing tools and libraries for generating authenticated REST requests to Amazon S3, to also send authenticated requests to Google Cloud Storage. The changes you need to make to your existing tools and libraries are described in this section.
To get set up for a simple migration do the following:
- Set a default Google project.
- Get a developer key.
In your existing tools or libraries, make the following changes:
- Change the request endpoint to use the Cloud Storage request endpoint.
- Replace the Amazon Web Services (AWS) access and secret key with the corresponding Cloud Storage access key and secret key (collectively called your Google developer key).
That's it! At this point you can start using your existing tools and libraries to send keyed-hash message authentication code (HMAC) requests to Google Cloud Storage.
When you use the Cloud Storage XML API in a simple migration scenario,
AWS signature identifier in the
Authorization header lets
Cloud Storage know to expect
x-amz-* headers and Amazon S3 ACL XML syntax in
Setting a default project
To use the Cloud Storage in a simple migration scenario, you must choose a default project. When you choose a default project, you are telling Cloud Storage the project to use for operations like GET service or PUT bucket.
To set a default project:
- Open the Cloud Storage Settings page in the Google Cloud Platform Console.
- Select Interoperability.
Click Make this PROJECT-ID my default project.
If the project is already the default project, you will see PROJECT-ID is your default project.
This project is now your default project. You can change your default project at any time by choosing a different project and enabling interoperable access for it.
Managing developer keys for a simple migration
To use the Cloud Storage XML API in a simple migration scenario, you must use keyed-hash message authentication code (HMAC) authentication with Cloud Storage developer keys. Developer keys consist of an access key and a secret. An access key is a 20 character alphanumeric string, which is linked to your Google account. All authenticated Cloud Storage requests, except those that use cookie-based authentication, must use an access key in the request so that the Cloud Storage system knows who is making the request. The following is an example of an access key:
A secret is a 40 character Base-64 encoded string that is linked to a specific access key. A secret is a preshared key that only you and the Cloud Storage system know. You must use your secret to sign all requests as part of the authentication process. The following is an example of a secret:
To generate a developer key:
- Open the Cloud Storage Settings page in the Google Cloud Platform Console.
- Select Interoperability.
- If you have not set up interoperability before, click Enable interoperability access.
- Click Create a new key.
Security tips for working with developer keys
You can create up to five developer keys. This is useful if you are working on several different projects and you want to use different developer keys for each project.
You can also use the key management tool to delete your developer keys and create new developer keys. You may want to do this if you think someone else is using your developer keys or you need to change your keys as part of a key rotation, which is a security best practice. If you have developer keys and you want to create new developer keys, we recommend that you first update your code with the new developer keys before you delete the old keys. When you delete developer keys, they become immediately invalid and they are not recoverable.
Authenticating in a simple migration scenario
For operations in a simple migration scenario that require authentication, you
will include an
Authorization request header just like you do for requests to
Amazon S3. The
Authorization header syntax for an Amazon S3 request is:
Authorization: AWS AWS-ACCESS-KEY:signature
In a simple migration scenario, you only change the header to use your Google Developer access key:
Authorization: AWS GOOG-ACCESS-KEY:signature
The parts of the
Authorization header are:
- Signature identifier
- The signature identifier identifies the signature algorithm and version that
you are using. Using
AWSindicates that you intend to send
- Access key
- The access key identifies the entity that is making and signing the request. In a simple migration, replace the Amazon Web Service (AWS) access key ID you use to access Amazon S3 with your Google developer access key. Your Google developer access key starts with "GOOG".
The signature is a keyed cryptographic hash of various request headers. The signature is created by using HMAC-SHA1 as the hash function and your secret as the cryptographic key. The resulting digest is then Base64 encoded. When Cloud Storage receives your signed request, it uses the access key to look up your secret and verify that you created the signature. For more information on how to obtain an access and secret key, see Managing developer keys for access in a simple migration scenario.
In a simple migration, replace the AWS secret access key with your Google developer key secret as the cryptographic key.
This section describes the process of authenticating an XML API request in a simple migration scenario. While this section can be used to develop your own code to sign requests, it is mainly intended to be a review if you already have tools or libraries that sign your requests to Amazon S3. In this case, you will continue to use these tools to access Google Cloud Storage using the XML API, with the changes shown here.
This authentication method provides both identity and strong authentication without revealing your secret. Providing both identity and authentication in every request helps ensure that every Cloud Storage request is processed under a specific user account and with the authority of that user account. This is possible because only you and the Cloud Storage system know your secret. When you make a request, the Cloud Storage system uses your secret to calculate the same signature for the request that you calculated when you made the request. If the signatures match, then the Cloud Storage system knows that only you could have made the request.
The following pseudocode shows how to create the signature for the Authorization header:
Signature = Base64-Encoding-Of(HMAC-SHA1(YourGoogleStorageSecretKey, MessageToBeSigned))
To create the signature, you use a cryptographic hash function known as HMAC-SHA1. HMAC-SHA1 is a hash-based message authentication code (MAC) and is described in RFC 2104. It requires two input parameters, both UTF-8 encoded: a key and a message. The key is your Cloud Storage secret and the message must be constructed by concatenating specific HTTP headers in a specific order. The following pseudocode shows how to construct the message:
MessageToBeSigned = UTF-8-Encoding-Of(CanonicalHeaders + CanonicalExtensionHeaders + CanonicalResource)
Each of the canonical entities that make up the message represents a strictly-ordered and strictly-formatted concatenation of various headers and resources. The following sections describe how to construct each of these entities.
You construct the
CanonicalHeaders portion of
concatenating several header values and adding a newline (U+000A) after each
header value. The following pseudocode notation shows you how to do this
(newlines are represented by
CanonicalHeaders = HTTP-Verb + "\n" + Content-MD5 + "\n" + Content-Type + "\n" + Date + "\n"
Do not include the header names in the concatenated string; include only the
header values. If a required header does not exist in a request, substitute an
empty string for the header value and be sure to include the newline after the
empty string. Also, the
Date header is required for all authenticated
requests, so this field must be populated with a valid date and time stamp. The
date and time stamp must be within 15 minutes of when Cloud Storage receives
You can also use the
x-goog-date extension headers to specify
the date and time stamp. If you use the date extension header, Cloud Storage
Date header in your request. In this case, substitute an empty
string for the
You construct the
CanonicalExtensionHeaders portion of the message by
concatenating all extension (custom) headers that begin with
x-goog-. However, you cannot perform a simple concatenation. You must
concatenate the headers using the following process:
- Make all custom header names lowercase.
- Sort all custom headers lexicographically by header name.
- Eliminate duplicate header names by creating one header name with a comma-separated list of values. Be sure there is no whitespace between the values and be sure that the order of the comma-separated list matches the order that the headers appear in your request. For more information, see RFC 7230 section 3.2.
- Replace any folding whitespace or newlines (CRLF or LF) with a single space. For more information about folding whitespace, see RFC 2822 section 2.2.3.
- Remove any whitespace around the colon that appears after the header name.
- Append a newline (U+000A) to each custom header.
- Concatenate all custom headers.
It's important to note that you use both the header name and the header value
when you construct the
CanonicalExtensionHeaders portion of the message. This
is different than the
CanonicalHeaders portion of the message, which used only
You construct the
CanonicalResource portion of the message by concatenating
the resource path (bucket, object, and subresource) that the request is acting
on. To do this, you can use the following process:
- Begin with an empty string.
- If the bucket name appears in the Host header, add a slash (/) and the bucket name to the string (for example, /travel-maps). If the bucket name appears in the path portion of the HTTP request, do nothing.
- Add the path portion of the HTTP request to the string, excluding any query string parameters. For example, if the path is /europe/france/paris.jpg?acl and you already added the bucket travel-maps to the string, then you need to add /europe/france/paris.jpg to the string.
- If the request is scoped to a subresource, such as ?acl, add this subresource to the string, including the question mark.
Copy the HTTP request path literally: that is, you should include all URL
encoding (percent signs) in the string that you create. Include only query
string parameters that designate subresources (such as
acl). You should not
include query string parameters such as
The following examples show what the signed message looks like for several different kinds of requests.
Sample authentication request
The following examples upload an object named /europe/france/paris.jpg to a
bucket named my-travel-maps, apply the predefined ACL
define a custom metadata header for reviewers. Here is the request to a bucket
in Amazon S3:
PUT europe/france/paris.jpg HTTP/1.1 Host: my-travel-maps.s3.amazonaws.com Date: Tue, 05 Nov 2013 23:46:19 GMT Content-Length: 888814 Content-Type: image/jpg x-amz-acl: public-read x-amz-meta-reviewer: joe,jane Authorization: AWS AWS-ACCESS-KEY:Signature
Here is the request for a bucket in Google Cloud Storage:
PUT europe/france/paris.jpg HTTP/1.1 Host: my-travel-maps.storage.googleapis.com Date: Tue, 05 Nov 2013 23:47:01 GMT Content-Length: 888814 Content-Type: image/jpg x-amz-acl: public-read x-amz-meta-reviewer: joe,jane Authorization: AWS GOOG-ACCESS-KEY:Signature
Here is the corresponding message to be signed for the Cloud Storage request,
PUT\n \n image/jpg\n Tue, 05 Nov 2013 23:47:01 GMT\n x-amz-acl:public-read\n x-amz-meta-reviewer:joe,jane\n /my-travel-maps/europe/france/paris.jpg
This request did not provide a Content-MD5 header, so an empty string is shown in the message (second line).
Access control in a simple migration scenario
To support simple migrations, Cloud Storage accepts ACLs produced by Amazon
S3. In a simple migration scenario, you will use
AWS as your
signature identifier which tells Cloud Storage to expect ACL syntax using
Amazon S3 ACL XML syntax. You should ensure that the Amazon S3 ACLs you use map
to the Cloud Storage ACL model. For example, if your tools and libraries use
Amazon S3's ACL syntax to grant bucket
WRITE permission, then they must also
READ permission because Cloud Storage permissions are
concentric. You do not need to specify both
permission when you grant
WRITE permission using the Cloud Storage syntax.
Google Cloud Storage supports Amazon S3 ACL syntax in the following scenarios:
- In a request to Cloud Storage to retrieve ACLs (for example, a GET Object or GET Bucket request), Cloud Storage returns Amazon S3 ACL syntax.
- In a request to Cloud Storage to apply ACLs (for example, a PUT Object or PUT Bucket request), Cloud Storage expects to receive Amazon S3 ACL syntax.
Authorization header in a simple migration scenario will use
AWS for the
signature identifier, but with your Google access key.
Authorization: AWS GOOG-ACCESS-KEY:signature
The following example shows a GET request to Google Cloud Storage to return the ACLs for an object.
GET europe/france/paris.jpg?acl HTTP/1.1 Host: my-travel-maps.storage.googleapis.com Date: Thu, 07 Nov 2013 01:04:05 GMT Content-Type: application/xml Authorization: AWS GOOG-ACCESS-KEY:Sku4/Nc1q4yJ6+7q7Sk3PFAAelE=
The response to the request includes the ACL using Amazon S3 ACL syntax.
<?xml version='1.0' encoding='UTF-8'?> <AccessControlPolicy> <Owner> <ID>00b4903a972faa8bcce9382686e9129676f1cd6e5def1f5663affc2ba4652490 </ID> <DisplayName>OwnerName</DisplayName> </Owner> <AccessControlList> <Grant> <Grantee xmlns:xsi='http://www.w3.org/2001/XMLSchema-instance' xsi:type='CanonicalUser'> <ID>00b4903a972faa8bcce9382686e9129676f1cd6e5def1f5663affc2ba4652490</ID> <DisplayName>UserName</DisplayName> </Grantee> <Permission>FULL_CONTROL</Permission> </Grant> </AccessControlList> </AccessControlPolicy>
The following example shows a PUT request to Google Cloud Storage to set the ACLs for an object. The example shows a request body with Amazon S3 ACL syntax.
PUT europe/france/paris.jpg?acl HTTP/1.1 Host: my-travel-maps.storage.googleapis.com Date: Thu, 07 Nov 2013 01:55:16 GMT Content-Type: application/xml Content-Length: 337 Authorization: AWS GOOG-ACCESS-KEY:Ugm1lh6Bb+nAfNVa5ljUgfXYMfQ= <?xml version='1.0' encoding='utf-8'?> <AccessControlPolicy> <AccessControlList> <Grant> <Grantee xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:type="AmazonCustomerByEmail"> <EmailAddress>firstname.lastname@example.org</EmailAddress> </Grantee> <Permission>FULL_CONTROL</Permission> </Grant> </AccessControlList> </AccessControlPolicy>
Finally, in a simple migration scenario, you can also use the
identifier in the
Authorization header. In this case, you must use the
Cloud Storage ACL syntax and ensure that all of your headers are Google
x-goog-*. While this is possible, we recommend that you move to a
full migration as described below in order to realize all the benefits of
Google Cloud Storage.
A full migration from Amazon S3 to Google Cloud Storage enables you to take advantage of all the features of Google Cloud Storage including:
- Support for service accounts
- Service accounts are useful for server-to-server interactions that do not require end-user involvement. For more information, see Service Accounts.
- Support for multiple projects
- Multiple projects allow you to have, in effect, many instances of the Google Cloud Storage service. This allows you to separate different functionality or services of your application or business as needed. For more information, see Using Projects.
- OAuth 2.0 authentication
- OAuth 2.0 relies on SSL for security instead of requiring your application to do cryptographic signing directly, and is easier to implement. With OAuth, your application can request access to data associated with a user's Google Account, and access can be scoped to several levels, including read-only, read-write, and full-control. For more information, see OAuth 2.0 Authentication.
To migrate fully from Amazon S3 to Google Cloud Storage, you will need to make the following changes:
- Change any existing
x-amz-*headers to corresponding
- Change AWS Access Control List (ACL) XML to the corresponding Cloud Storage ACL XML (see Specifying Bucket and Object ACLs).
- Set the x-goog-project-id header in your requests. Note that in a simple migration scenario, you chose a default project for all requests. This is not needed in a full migration.
Get set up to use OAuth 2.0 authentication as described in OAuth 2.0 Authentication. The first step is to register your application (where you will be issuing requests from) with Google. Using OAuth 2.0 means that your
Authorizationheader will look like this:
Authorization: Bearer <oauth2_token>
Access control in a full migration
This section shows a few examples of access control to help you migrate from Amazon S3 to Google Cloud Storage. For an overview of access control in Google Cloud Storage, see Access Control.
In Google Cloud Storage, there are several ways to apply ACLs to buckets and objects (see Specifying Bucket and Object ACLs). Two of the ways you specify ACLs are analogous to what you do in Amazon S3:
aclquery string parameter to apply ACLs for specific scopes.
x-goog-aclrequest header lets you apply predefined ACLs, which are sometimes known as canned ACLs.
Using the acl query string parameter
You can use the
acl query string parameter for a Cloud Storage request
exactly the same way you would use it for an Amazon S3 request. The
parameter is used in conjunction with the PUT method to apply ACLs to the
following: an existing object, an existing bucket, or a bucket you are creating.
When you use the
acl query string parameter in a PUT request, you must attach
an XML document (using Google Cloud Storage ACL syntax) to the body of your request. The
XML document contains the individual ACL entries that you want to apply to the
bucket or object.
The following example shows a PUT request to Amazon S3 that uses the
string parameter. ACLs are defined in an XML document sent in the request body.
The PUT request changes the ACLs on an object named europe/france/paris.jpg
that is in a bucket named my-travel-maps. The ACL grants email@example.com
PUT europe/france/paris.jpg?acl HTTP/1.1 Host: my-travel-maps.s3.amazonaws.com Date: Wed, 06 Nov 2013 19:28:18 GMT Content-Length: 598 Content-Type: application/xml Authorization: AWS AWS-ACCESS-KEY:eNB6IUnqzsVAZ69TZZ5yFMQ5SvE= <?xml version='1.0' encoding='utf-8'?> <AccessControlPolicy> <Owner> <ID>5a6557ba40f7c86496ffceae789fcd888abc1b62a7149873a0fe12c0f60a7d95</ID> <DisplayName>ownerEmail@example.com</DisplayName> </Owner> <AccessControlList> <Grant> <Grantee xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:type="CanonicalUser"> <ID>fd447671d60b979f78ee6fcec7b22afc80e6b26a4db16eed01afb8064047949b</ID> <DisplayName>firstname.lastname@example.org</DisplayName> </Grantee> <Permission>FULL_CONTROL</Permission> </Grant> </AccessControlList> </AccessControlPolicy>
Here is the same request to Google Cloud Storage:
PUT europe/france/paris.jpg?acl HTTP/1.1 Host: my-travel-maps.storage.googleapis.com Date: Wed, 06 Nov 2013 19:37:33 GMT Content-Length: 268 Content-Type: application/xml Authorization: Bearer ya29.AHES6ZRVmB7fkLtd1XTmq6mo0S1wqZZi3-Lh_s-6Uw7p8vtgSwg <?xml version='1.0' encoding='utf-8'?> <AccessControlList> <Entries> <Entry> <Permission>FULL_CONTROL</Permission> <Scope type="UserByEmail"> <EmailAddress>email@example.com</EmailAddress> </Scope> </Entry> </Entries> </AccessControlList>
Note that Cloud Storage does not require an
<Owner/> element in the
ACL XML document. For more information, see Default Object ACLs.
You can also retrieve bucket and object ACLs by using the
acl query string
parameter with the GET method. The ACLs are described in an XML document, which
is attached to the body of the response. You must have
to apply or retrieve ACLs on an object or bucket.
Applying ACLs with an extension request header
You can use the
x-goog-acl header in a Cloud Storage request to apply
predefined ACLs to buckets and objects exactly the same way you would use the
x-amz-acl header in an Amazon S3 request. You typically use the
x-amz-acl) header to apply a predefined ACL to a bucket or object when you
are creating or uploading the bucket or object. The Cloud Storage predefined
ACLs are similar to Amazon S3 Canned ACLs, including private,
public-read, public-read-write, as well as others. For a list of Cloud Storage
predefined ACLs, see Predefined ACLs.
The following example shows a PUT Object request that applies the
ACL to an object named europe/france/paris.jpg that is being uploaded into a
bucket named my-travel-maps in Amazon S3.
PUT europe/france/paris.jpg HTTP/1.1 Host: my-travel-maps.s3.amazonaws.com Date: Wed, 06 Nov 2013 20:48:42 GMT Content-Length: 888814 Content-Type: image/jpg x-amz-acl: public-read Authorization: AWS AWS-ACCESS-KEY:o8PRLyrTxsMEznHzgQvTt2TnWTI= <888814 bytes in entity body>
Here is the same request to Google Cloud Storage:
PUT europe/france/paris.jpg HTTP/1.1 Host: my-travel-maps.storage.googleapis.com Date: Wed, 06 Nov 2013 20:49:57 GMT Content-Length: 888814 Content-Type: image/jpg x-goog-acl: public-read Authorization: Bearer ya29.AHES6ZRVmB7fkLtd1XTmq6mo0S1wqZZi3-Lh_s-6Uw7p8vtgSwg <888814 bytes in entity body>
You can also use the
x-goog-acl header to apply a predefined ACL to an
existing bucket or object. To do this, include the
acl query string parameter
in your request but do not include an XML document in your request. Applying a
predefined ACL to an existing object or bucket is useful if you want to change
from one predefined ACL to another, or you want to update custom ACLs to a
predefined ACL. For example, the following PUT Object request applies the
private to an object named europe/france/paris.jpg that is
in a bucket named my-travel-maps.
PUT europe/france/paris.jpg?acl HTTP/1.1 Host: my-travel-maps.storage.googleapis.com Date: Wed, 06 Nov 2013 00:26:36 GMT Content-Length: 0 x-goog-acl: private Authorization: Bearer ya29.AHES6ZRVmB7fkLtd1XTmq6mo0S1wqZZi3-Lh_s-6Uw7p8vtgSwg <empty entity body>
For more information about managing ACLs, see Managing Access Control.
Migrating from Amazon S3 to Cloud Storage Request Methods
Google Cloud Storage supports the same standard HTTP request methods for reading and writing data to your buckets as are supported in Amazon S3. Therefore, the majority of your tools and libraries that you currently use with Amazon S3, will work as is with Google Cloud Storage. Cloud Storage supports the following request methods:
- Service request for GET.
- Bucket requests, including PUT, GET, DELETE.
- Object requests, including GET, POST, PUT, HEAD, and DELETE.
For more information, see XML API Reference Methods. Keep in mind that when you send requests to Google Cloud Storage, you will need to change the request body, when applicable, to use the appropriate Google Cloud Storage syntax. For example, when you create a lifecycle configuration for a bucket, use the Cloud Storage lifecycle XML, which is different than the Amazon S3 lifecycle XML.
There are a few differences between Google Cloud Storage XML API and Amazon S3 which are summarized below, with suggested Cloud Storage alternatives:
|Amazon S3 Functionality||Cloud Storage XML API Functionality|
POST /<object-name>, PUT /<object-name>
In the Google Cloud Storage XML API, you can upload a series of component objects, performing a separate upload for each component. Then you can compose the objects into a single composite object.
Note: While the JSON API offers a multipart upload feature, this feature is used for sending metadata along with object data. It is not equivalent to S3's multipart upload feature.
|GET/POST bucket query string parameters:
|Multiple object delete.
Use the gsutil rm command to easily remove multiple objects. The rm command supports the "-m" option to perform parallel (multi-threaded/multi-processing) deletes.
Alternatively, the JSON API supports sending batch requests to reduce the number of HTTP connections your client makes.
Migrating from Amazon S3 to Cloud Storage Headers
Google Cloud Storage uses several standard HTTP headers as well as several custom (extension) HTTP headers. If you are transitioning from Amazon S3 to Google Cloud Storage, you can convert your custom Amazon S3 headers to the equivalent Cloud Storage custom header or similar functionality as shown in the table below.
|Amazon S3 Header||Cloud Storage Header|
||Not required. Google Cloud Storage automatically encrypts all data before it is written to disk. For more information, see Encryption.|
||You can specify storage class when you create a bucket. For more information, see Storage Classes.|
||Use OAuth 2.0 Authentication.|
Discussion groups and support for XML API compatibility with Amazon S3
The Google Cloud Storage gs-discussion group which formerly supported XML API interoperability and migration issues, is in archive mode. The discussion forum can now be accessed on Stack Overflow using the tag google-cloud-storage. See the Resources and Support page for more information about discussion forums and subscribing to announcements.