Object Storage provides the ability to turn any file and any associated metadata into an object accessible by an HTTP API from anywhere. It ensures data integrity, provides data protection, and scales to millions of objects without requiring traditional filesystem management. Object Storage excels at handling unstructured data; and provides a location for you to store objects to be accessed by instances or if you wish generally online as well.
Object Storage doesn’t have support for a deep hierarchy; objects are stored in containers but containers may not be inside another container. It’s possible to mimic folders by using ‘/’ in an object’s name but from a system point of view there is no folder support.
Our implementation of object storage is powered by OpenStack’s Swift.
Some Use Cases
Static Web Assets
Large amounts of unstructured data (eg. map tiles, photos, etc.)
Archival (eg. tax records, backups, etc.)
Apps that leverage extensive metadata on a file
You can gain access to your containers and objects using several methods including the RAC Dashboard, the swift command line client, interacting with the Swift API itself, or using a popular GUI application. Of note at present we don’t offer the S3 compatibility API used by some libraries to connect to Swift. (If you would like to use these feature please contact us at firstname.lastname@example.org)
You can access your objects from anywhere in the world - whether it's on one or more of your instances (eg. a shared storage alternative), your personal computer, or elsewhere. By default only you (when provided with your credentials) can access your objects, and with ACLs you can manage read and write access for other users on a container level.
We offer Swift in both the Calgary region and the Edmonton region.
You can create, delete, copy, and download objects and containers using the RAC Dashboard. Additional features such as setting metadata which is used for most advanced features are not available via the Dashboard.
Command Line (swift client)
If you haven’t already, it’s highly recommended you have reviewed Command-line Tools in the Advanced Guide. Installing the
python-openstackclient should include installation of the
python-swiftclient, but if you need to manually install it, follow the instructions below.
After sourcing your openrc file you can then use the swift command to interact with the object storage. Examples can be found below.
Additionally your application can leverage the Swift API directly - either using curl or existing libraries (such as Boto for Python).
Our OpenStack Swift environment has support for S3 API compatibility. This means that any application or library which supports Amazon S3 should be able to be used in our Swift environment.
To begin using the S3 API, you first need to obtain a set of "EC2 Credentials". This is a set of credentials formatted in the same way as Amazon's credentials. You can download your EC2 Credentials by going here (https://cloud.cybera.ca/project/api_access/) and clicking "Download EC2 Credentials". This will trigger a zip file to be downloaded to your workstation.
Unzip the file and look inside the
ec2rc.sh file using a text editor of your choice. Inside this file, you will see a bunch of variables, including:
These are your EC2 credentials.
Most Amazon applications and libraries use
AWS_ACCESS_KEY_ID instead of
AWS_SECRET_ACCESS_KEY instead of
Cyberduck is available for Mac OS X and Windows.
To use Cyberaduck using the standard S3 compatibility, first obtain your EC2 Credentials, as described in the S3 API of this document.
Open Cyberduck and create a new Amazon S3 connection. Fill in the following information:
yyc.cloud.cybera.ca(for the Calgary region) or
yeg.cloud.cybera.ca(for the Edmonton region).
- Port: 8080
- Access Key ID: Your
- Secret Access Key: Your
OpenStack Swift Native
To use Cyberduck with the Swift API, you’ll need to download a special profile to set some extra settings for bookmarks in Cyberduck. You can download the generic profile or a profile we have created that is specific to RAC.
1. Download the additional profile from Cyberduck’s website called OpenStack Swift (Keystone). Double click on this file and enter the following details in a new bookmark using the OpenStack Swift (Keystone) preset to access the object storage:
This method does have the occasional issue where you need reset the port number (it will also display a warning about sending the password as clear text)
DO NOT upload files larger than 2 GB using Cyberduck if you wish to share them via the Web.
In Depth Features / Tutorial
See Swift CLI reference for more information.
Create a Container
Upload a File
Add and Check metadata on a file
Download a File
The second project may then interact with the shared container by appending the storage-url for the container to their swift commands:
Making Files Available via the Web
Set container to be public to anyone
Figure out URL: by finding the AUTH_xxxx string by running
eg. http://swift.cloud.cybera.ca/v1/AUTH_xxxx/filename.txt or by using Cyberduck (see above) and choose Copy URL.
Alternatively you can also enable web listings so you can see all the objects in a container:
And then visit http://swift.cloud.cybera.ca/v1/AUTH_xxxx/container_name to view all the objects (changing AUTH_xxxx for your AUTH information)
Another feature Swift offers is automatic versioning. This means every time an object is uploaded or updated it will store the old version in a special versions container that is created.
This means the older versions are still all accessible in a second container. You can enable this by setting the ‘X-Versions-Location’ header on your container:
One last feature we wanted to highlight is the ability to offer temporary URLs for objects or expiring URLs as they are also called. This allows you to provide a URL that will stop functioning after a certain amount of time.
To do this you need to be able to set a secret key in a header on your account to allow Temporary URLs, and then create an HMAC-SHA1 signature generated from the HTTP method, the expiration timestamp, the path to the object, and the secret key set in the header before. It’s not as complicated as it sounds however.
The first step is to create the secret key for your account:
Then we need to create our temporary URL. You can use a tool such as swift-temp-url (Python), or have your application create the link by creating a signature based on the method, expiration timestamp, and the path. The above linked tool uses the following code to do this:
To use swift-temp-url - download the file to your computer, and set the file to be executable. You can then run it and get your URL. eg. a link that lasts for 10 minutes
Results in a shareable link:
Be sure to set set your ACCOUNT ID, and then the path (proper container name and object name)
Large/Segmented Object Warnings
Swift segments files to support objects larger than 5 GB. If you want to upload a file larger than 5GB using the Swift CLI you will need to ensure you use the -S flag and set a segment size.
By default Swift splits files larger than 5 GB while Cyberduck performs this for if the file is larger than 2GB. There are a couple caveats with the different approaches the two applications take to handling large files when it comes to accessing these files via a web browser (completely unrelated to the size they use to split files).
Object uploaded via Swift CLI:
Swift will create a <containername_segments> container that holds the separate files. You’ll need to enable read access AND rlistings for this container in order for the download to work. For a container named public you’ll need to make sure the following is set:
Once both are set, downloading the object via your web browser as an unauthenticated user will work as expected. (Bug #1082835)
Object uploaded via Cyberduck:
Cyberduck creates a folder called .file_segments in your container and places the segments in there. It then writes JSON data to the manifest file so you can download the file later
Container sync offers the ability to synchronize the contents of two or more containers between the two regions of the Rapid Access Cloud or if supported, other OpenStack Swift installations. Every five minutes, a sync will run copying any new or changed metadata and objects (including deletions) from the source container to the destination container. Depending on the size of the objects it may take a couple more minutes to copy the object in it's entirety.
This is done by creating a sync relationship between any two swift containers. In the example below, we will create a container in Calgary (container1) and another in Edmonton (container2). Each container will sync to the other (two-way replication) which is useful if you want the objects to be highly available. We will need a key for each container to share for security (the key will simply be ‘secret’, but for production please use a complex password).
Determine your AUTH ID (Account ID, line 3):
Create or modify a container in Calgary and include the `--sync-to` and `-k` attributes (the `swift post` command updates info for a container, or creates one if it doesn’t exist):
Create or modify a container in Edmonton to sync to Calgary:
It is possible to do a one-way sync for disaster recovery, effectively creating a backup container; in our example above, if we wanted to sync one-way from yyc => yeg we create the container in Calgary the same, but the container in Edmonton will not have a `--sync-to` switch, only a `-k`:
Now objects in Calgary will replicate to Edmonton, but not the other way.
Disabling Container Sync
At present disabling container sync is not available in the
openstack command line tool at the time of writing. You must use the
One example of why you might want to do this if you previously had a two-way container sync with YEG ↔ YYC and only wish to sync one-way now.
To remove the sync setting, provide an empty
--sync-to parameter instead versus filling in the parameter to set up sync as seen in the section above.
If you'd like to disable the sync entirely, run the command on each container in both regions.