Google Cloud Platform (GCP) vs. Amazon Web Services (AWS) for the Hobbyist
A look at the free tiers on these platforms and how they compare.
A good chunk of my free time is spent tinkering, designing, and implementing systems that some would likely describe as “overengineered”. Do I really need a distributed Pub/Sub platform and a Data Lake for a home automation platform with 3 devices producing less than a kilobyte of data daily that gets consumed by a single web application? Probably not… but how else am I going to find the opportunity to figure out how these types of cloud services integrate and complement each other?
This post is not about convincing you to adapt my approach to your personal hobbies but instead to share with you why I chose GCP over AWS for my personal projects and maybe help make this decision easier for you. I won’t go into great detail about comparing the product interface and features. My goal is to explain what the core products are in the trial and free tiers, how they can be used, and how their trials compare.
Money is no object?
Read no further, if your hobby does not have budget restrictions there can be no question, AWS is a clear choice. The overall offering is more complete and robust.
So which platform has the best free stuff?
With that out of the way let’s take a look at what core free features the two cloud providers make available:
The lists of free and trial products on GCP and AWS are vast and I tried to cover those that would be of most value to hobbyist developers such as myself. You can check out the full lists here for GCP and here for AWS. Let’s go through the options listed in my table, understand the product availability, and their application for potential projects.
Compute
GCP offers an always free Computer Instance while AWS offers Elastic Compute Cloud (EC2). There is no question that with the free option AWS provides more powerful servers in this bracket but unfortunately, after the 12 monthly trials, AWS will be charging you for any uptime. GCP on the other hand has pledged to provide you with an always free instance of the bottom tier of CI.
What I use it for: The instances are private virtual machines and can be used for various purposes from web applications to the hosting of databases and other services.
Verdict: For me, GCP wins easily in this category as I can continue to leverage the service free of charge indefinitely.
NoSQL Database
Both platforms offer fairly robust NoSQL options free of charge up to a certain amount of storage and transactions. GCP offers the Firestore database with 1GB of free storage and AWS offers DynamoDB with 25GB of free storage.
What I use it for: These databases are best used for storing data that does not require complex relational structures (e.g. user table).
Verdict: Unless you are building a large production level platform, 1GB of NoSQL storage should be more than enough for your needs but in just this category I would say AWS does come out on top.
Serverless Compute
Again both platforms provide always free tiers with the GCP Cloud Functions product allowing up to 2 million free requests per month and the AWS Lambda offering up to 1 million.
What I use it for: The list of uses for serverless compute is extremely long but I mostly utilize it for processing events in real-time and batching jobs.
Verdict: Just as with the NoSQL option the free quota should more than cover your needs in either platform unless you are building extremely heavy-duty platforms so I would say the two are fairly matched.
Serverless App Hosting
GCP App Engine always free tier provides 28 hours of hosting per day, AWS does provide a Lightsail trial in this category but only on a 30-day trial.
What I use it for: These platforms are ideal for hosting web applications such as Angular/React projects or Data APIs over your data sources.
Verdict: For me, there is no question about it, GCP wins flat out due to the free tier option. If you get creative you can squeeze in 2 applications @ 14 hours a day.
Serverless API
GCP Cloud Run provides container execution services with an always free 2 million requests per month, AWS does have the Fargate product but there is no free version available.
What I use it for: These container execution instances are perfect for building API layers on top of databases or serving machine learning modules.
Verdict: GCP wins as the service on their platform is always free!
Stream Processing
GCP Pub/Sub provides an event streaming platform with 10GB of monthly data processing in the always free tier. AWS has many different offerings in this category including Managed Kafka and Kinesis but all these options are paid and have no free offering.
What I use it for: I mostly leverage these platforms for sharing data between multiple systems.
Verdict: GCP wins as the service on their platform is always free!
File Storage (Data Lake)
GCP Cloud Storage provides 5GB of always free storage and AWS provides the same 5GB of storage via a 12 month trial on their S3 platform. I do prefer the S3 API to the one provided by GCP CS however for the main purpose of serving as a file store both products do just fine.
What I use it for: There are many uses for buckets in these file stores, from serving public resources (images, JavaScript files, etc.) to blob storage for unstructured data, to the storage of project files.
Verdict: GCP wins as the service on their platform is always free!
SQL Database
AWS provides an RDS t2.micro instance for 12 months for free where you can host one of many different SQL databases. GCP does provide Cloud SQL for this purpose but it has no free tier.
What I use it for: Storage of relational data.
Verdict: AWS wins as the service on their platform is free for 12 months!
Conclusion
At the end of the day, it really comes down to what functionality you are really looking for and your time frame. I have been tinkering for years and my 12-month trials on AWS are long gone. GCP offers enough hosting, data processing, and storage for free to more than cover my multiple hobby projects.
