Hi everyone, author here. I wrote this article to summarize my thoughts on using the Google Cloud vs Amazon.

TL;DR: Although AWS has a lot more cloud products, unless you need the additional options, IMHO the Google Cloud is more intuitive, cheaper and offers better cost structure (e.g. by-the-minute pricing instead of by-the-hour).

So if you’re starting a new project, I would highly recommend that you give the Google Cloud a try.

Yo! Totally legitimate question, as I don't know the answer to this, and I'm extremely curious: I see a lot of people say that Google is cheaper than Amazon, but I also see a lot of people bemoaning reserved instances and buried in every comment thread about Google vs. Amazon someone points out "wait, this comparison isn't taking into account reserved instances". I thereby don't feel like I actually know which offering is cheaper (though I'm betting it is Google, mostly for the same reasons I don't really want to use their offering, which is maybe unfortunate: that I hate free tiers and artificially low prices, as they either turn into a tax somewhere else or a reason to later scrap a product line). Can you verify that your comparison with 25% cheaper is based on actually paying attention to the pricing structure of the two products? What I'd like to see is "assuming you are going to use a computer for two or three years, and are willing to spend the twenty minutes to save potentially thousands of dollars (as if you aren't, something is wrong with your priorities ;P <- note that I'm willing to admit my priorities are also broken :/...), here is the cost difference between these two offerings".

Great question!

Until recently, Google Cloud didn't have reserved instances. Interestingly enough, last week they introduced their version of it called "Committed Use" pricing.

Here's what it looks like for 2 CPUs/8GB RAM (numbers below according to their documentation page):

  AWS
  On-demand = $69/month
  1-year    = $53/month
  3-year    = $47/month

  GCP
  On-demand = $52/month
  1-year    = $31/month
  3-year    = $22/month

So GCP is cheaper even considering the reserved instances!

Are you talking about this article? https://thehftguy.com/2016/11/18/google-cloud-is-50-cheaper-...

In short. If your infrastructure is so small that it's about reserving 3 instances for a year. You should be running on Digital Ocean. It's much more simple and cheaper. You likely don't need the more advanced features from Google/Amazon.

If your infrastructure is in the hundred of VMs. Reserved instances are basically unusable, it's a NP complete problem to optimize capacity for a moving target. You're as likely to save 10% as to lose 20% on your reservations.

Last but not least. Reserved instances have many tricks and constraints that you'll only find out after you spend 10k on them, and can't go back.

We actually use Google's Firebase to manage our Alexa skill, delivering content from S3, best of both worlds.

That's a really interesting approach, thanks for sharing!

This blog post really just compares EC2 vs GCE. That's a valuable comparison to make, but covers a fairly narrow use case. I'd love to see comparisons that that say "for EDW X wins over Y", "for high volume message passing Z is best", "for data archival Q wins" etc

If price is your thing, Linode for compute and Backblaze for storage; 4 CPU and 8GB RAM for $40 a month and 1/2 cent per month per gig. For full feature cloud I find Azure a better competitor to AWS as far as features/price/interface and per minute billing.

well only linode if security is not your thing... They have a history of security issues.

I think this is the latest in a long line: https://www.theregister.co.uk/2016/02/09/linode_ssh_security...

Nice article. Good explanation of what some of the major numbers and features mean in a practical sense.

Basically, an advertisement for Google barely disguised as an 'article'.

Sorry you feel that way, are there any points you disagree with?

I'm in no way affiliated with or remunerated by Google. I've just spent a lot of time in the past comparing both and wanted to share that with the community.