Should I choose Amazon or Google for cloud services?

Google Cloud wins on price

In terms of compute and storage costs, Google Cloud is the clear winner. For example, 2 CPUs / 8GB RAM costs $69 per month on AWS, while GCP is only $52 (25% cheaper). And for cloud storage costs, GCP's regional storage costs are only 2 cents/GB per month and AWS 2.3 cents/GB per month. Additionally, GCP offers a "multi-region" cloud storage option, where data is automatically replicated across multiple regions, for 2.6 cents/GB/month. (Detailed calculations can be done through AWS costing and GCP costing.)

Besides the cheap price, GCP also offers a better cost structure. Instead, AWS offers a pay-per-hour model, and GCP offers a pay-per-minute model, starting from 10 minutes. This is important if using the cloud, where the startup time is relatively short, because AWS is incrementally calculated on a whole-hour basis, which adds to the cost. For example, an analysis job takes 2.01 hours, GCP will charge the calculation fee for 2.01 hours, and AWS will charge it for 3 hours aws vs azure vs google cloud.

In addition, GCP offers better discounts for long-term users: unlike AWS, which requires users to pre-pay for long-term services, GCP automatically discounts you for long-term use and does not require advance reservations. If reserved customizations suit your needs, buy reserved customizations at GCP now for an extra discount aws vs azure vs google cloud!

AWS wins in market share and services

In terms of user adoption, AWS is the undisputed market leader, accounting for almost half of the infrastructure-as-a-service (IaaS) market. With its first-mover advantage and nearly 5 years of development, AWS has more cloud products and options. In contrast, GCP is new to the field and offers similar solutions, but is still a bit behind aws vs azure vs google cloud.

For example, if you need a fully managed cloud SQL solution, GCP offers a managed MySQL solution (PostgreSQL was released in beta last week), while AWS also offers options like Aurora, MariaDB, Oracle, and Microsoft SQL Server . In addition, AWS offers a "serverless" computing product called AWS Lambda that allows users to run code at any time without having to set up an instance dedicated to accepting requests. While GCP offers a similar product (Google Cloud Functions), it's only just released in beta, so it's something to think about before using GCP in practice aws vs azure vs google cloud.

In terms of global accessibility, AWS has more datacenters all over the place. If you do business in China, keep this in mind: you can access GCP instances in China, but not (except Hong Kong) files stored in Google Cloud Storage (the equivalent of Amazon S3) aws vs azure vs google cloud.

Google Cloud wins in instance configuration

When it comes to large scale instances, AWS is the clear winner, the largest GCP instance is 64 CPUs / 416 GB RAM, while AWS offers instances with 128 CPUs and up to 2TB of memory!

That said, GCP is more flexible in instance configuration. In addition to having similar predefined instance types like AWS, GCP allows users to customize the amount of CPU and memory to use. For example, the instance type N1-STANDARD-1 is equipped with 1 CPU and 3.75GB of memory, but you can choose to have 1 CPU and 1.75GB of RAM, 4.25GB or 5GB of memory. If your computing needs are between the two instances, custom machine types can significantly reduce costs.

Second, if you need a lot of on-the-fly analytics that can easily be done in small chunks, then you'll need to understand which cloud solutions offer cheaper ephemeral instances. You may be familiar with AWS's spot instance (spot instance), buying instances through spot bidding (generally much cheaper than non-spot), but if the market price exceeds your bid, you will lose the eligibility to continue to use the instance (AWS also provides If the bidding block is set, the time can be set in advance). GCP also has a similar service, but it does not adopt the bidding model. Their service is called preemptible instance. Such machines can run for up to 24 hours, but Google can terminate their operation at any time if these computing resources are needed. When the instance is preempted, GCP will run your predefined shutdown script, which will save 30 seconds of current state analysis time. The advantage of the no-bid mode is that the startup process of preempting an instance can be easily automated, and the price is more predictable, with the lowest price being 20% ​​off the regular price!

Google Cloud wins for free

AWS is very generous in offering a one-year free trial, which is long enough. A one-year free trial includes 750 hours/month for a small 1 CPU/1GB RAM instance, 30GB disk storage, 750 hours/month for a similar size Managed database instance (e.g. MySQL) and 5GB cloud storage (enough for a small web server to run uninterrupted for a full year). However, there are more perks: AWS's free trial also offers many other free products, so be sure to check out the full list on the AWS website.

Until last week, Google Cloud was only offering a 60-day trial for a total of $300, which didn't look like a trial but more like a $300 discount. As of now, GCP has extended its $300 limit by 12 months and added a free tier with no time limit. For example, you can get a 0.2 CPU/0.6 GB RAM instance with 30 GB disk storage and 5 GB cloud storage for free. If this stays the same, then you can run a small website on GCP for free and run forever (but that's another topic). Please visit the GCP website for details.

Another reason GCP's trial wins is that the "credit" model is better for users new to cloud computing, as it forces you to think about how much you're spending during your trial. In my experience, this saves you from getting a staggering bill after the trial period ends.

Summarize

In conclusion, AWS does offer more cloud offerings, but frankly, unless your application really needs them, I find it more appropriate to use a "less is more" mentality when it comes to cloud options. For people who are new to the cloud and don't know what to choose, offering a gentler learning curve is more welcome.

In my experience, Google Cloud's intuitive interface, coupled with cheaper costs, flexible billing options, per-minute pricing models, and shootable instances, makes Google Cloud Platform a very strong competitor to AWS.

So, if you are starting a new project, I would highly recommend that you definitely try Google's cloud platform.