Choosing between the three biggest cloud providers - Amazon Web Services (AWS), Microsoft Azure and Google Cloud Platform (GCP) - can be a testing proposition, but one of the best things about them is that they all offer a free tier for spinning up an instance and trying out the platform before you buy.
Companies are increasingly turning to multiple cloud vendors to avoid lock-in with one of the big three, or to take advantage of unique capabilities within each platform, especially when it comes to AI and machine learning.
That’s why free tiers are vital to organisations looking to test out capabilities before diving in and providing their credit card details, with all three offering various ways to consume their cloud services for free, whether through credits or by applying tight usage limits.
AWS had great success in its early days with this model, proving the value of outsourcing your technology infrastructure - also known as infrastructure-as-a-service (IaaS). Now the two leading rivals are increasingly looking to attract developers to their cloud platforms with credits and free access to their growing range of cloud tools and services.
So, who offers the most generous free tier today?
Google Cloud Platform
Google Cloud Platform announced an ‘always free’ tier in March 2017 for any organisations with modest usage needs, perfect for prototyping or private betas, as well as its old policy of offering credits for new users: $300 for the first 12 months to be exact, with no auto charge kicking in after the trial.
Always free offers up to 1GB of Google Cloud Datastore capacity, 28 instance hours per day for Google App Engine, one micro sentence per month for Google Compute Engine, 5 GB-months of Google Cloud Storage (regional only), 2 million Cloud Functions per month, 50GB of logs with Stackdriver for monitoring, as well as limited access to products like: Google Cloud Natural Language, Cloud Vision API, Kubernetes Engine and more.
Microsoft Azure now offers a similar model but with fewer credits for new sign ups.
With an Azure free account you get £150 credit to explore services over 30 days, followed by 12 months of access to those services for free, with limits.
This includes 750 hours of Windows or Linux Virtual Machines for compute, 250GB of SQL database storage, 5GB of Blob or simple storage, 1 million functions a month, and 15GB of outbound networking bandwidth, as well as the always free services below.
Microsoft now also offers always free too, giving access to more than 25 Azure services for free year-round. However this doesn’t include core services like compute and storage, instead allowing limited access to more niche services like Bing Speech, Face API, machine learning studio, IoT Hub and more.
AWS still offers credits for students and startups, and has a free tier with similar limits to GCP, which is limited to 12 months, and an ‘always free’ tier which is more limited and doesn’t include core products like S3 storage and EC2 (elastic compute).
The 12 month option offers 1 million API calls per month, 750 hours a month of EC2, 5GB of S3 storage, 30GB of Elastic Block storage, 500MB of Elastic Container Registry, and access to machine learning products like: Lex, Polly, Rekognition, Translate and Transcribe.
The always free offering from AWS is more aimed at getting developers acquainted with developer tools like CodeCommit and X-Ray or CloudWatch monitoring, as well as 1 million Lambda functions a month and Glacier object storage.
As with any comparison like this it very much depends on what you are looking to do and a fair amount of personal preference.
Google certainly offers some of the more generous limits to allow developers to get some serious prototyping done, and it’s always free tier is by far the most full featured of the three. On the flip side, Azure looks the stingiest.
However Azure and AWS have their own unique products that users may want to get their hands on, like Lex for building voice interfaces, or Microsoft’s Face API for facial recognition, that’s not to say that Google hasn’t got machine learning expertise of its own.
Luckily all you need to do to get started is create an account.