Google is following AWS ,. Microsoft is also following AWS . Which is not necessarily bad but doesn’t speak much of their innovation . The only play they have against AWS is fear mongering about vendor lock-in which is highly debatable anyways .
Oh, give me a break... where did Kubernetes come from? Whose LB solution relies on DNS and whose uses BGP/ECMP for regional routing and failover? Who offered HIPPA compliance without dedicated hosts and stupid markups first? Who created BigQuery?
Look, both Google and AWS have areas where each excels over the other, but to say that Google isn’t innovating is to say you are basically ignorant of their offering.
Google's pricing is also a lot more attractive. Unlike Amazon, they don't cut you from the free tier after a year. For a small toy project that's all you need.
To elaborate more, automatic discounts are really a big deal even at scale because it’s somewhat similar to AWS pricing schemes for reserved instances except it doesn’t require you to reprovision anything at all. For a lot of companies with stateful software and non-cloud native architecture (lift and shift) this helps cut significant costs while you’re growing.
The default pricing mechanism in AWS for on-demand instances in particular for companies that can’t afford to rewrite their non-cloud native software with tons of state and no CI/CD infrastructure or process is a boon for AWS and even at six figure monthly spend you could certainly use a 40%+ discount on your instances.
NLB DynamoDB EMR even HIPAA without dedicated was on AWS first ... whatever AWS does Google is trying to do better. Granted Kubernetes was google product but what was so innovative about it when Docker and Linux container was already there for a long time ... and now AWS has ECS
You are factually incorrect about HIPPA and dedicated hosts. Amazon only switched to the non-dedicated approach this year some time. Google has been that way for at least two (if not longer).
EMR is basically hosted Hadoop, which is an open source implementation of a system Google created and made famous.
Docker is based on cgroups, which was largely contributed by Google engineers. And anyone deploying docker for distributed systems will tell you that you need some kind of scheduler/orchestrator.
Amazon excels at packaging and bundling the work of others, and provides a fairly robust enterprise management layer on top of it. CloudFormation, for example, is much better than GCP’s Deployment Manager. Google is quickly catching up in the enterprise management space as well, as they take on larger customers who demand these abilities, but I still feel like AWS has a bit more capability there.
You probably should have listed Lambda, which was a nice idea out of AMZN.
Vendor lock-in is most certainly not the only competing point we have against AWS. From BigQuery, to BigTable, to Spanner, to our private network, to our live migration of compute, to our hosted Kubernetes, the list goes on and on.
AWS does a lot of things right, and we're more than happy to go toe-to-toe with them in the areas they have us beat. Competition is a sign of a healthy economy where everyone benefits. How can you compete if you don't even try to match the people you are competing against?
Speaking of innovation, I think google's innovation in cloud has started long before AWS. Many things that's ubiquitous today was google's innovation that's been battle tested internally for serving google ad and search business. Google has been a bit late to cloud game, but i guess it's going strong.
Disclaimer: Google employee, not working on cloud stuff, but admire how cloud has evolved over time.
Google is quickly moving to containerized solutions for almost everything you might want to deploy.
Avoiding lock-in, where reasonable (you're never going to be able to carry out something like big query as far as I can tell, and I don't work on those products or have special knowledge there) is something close to our hearts and is an important topic during design discussions of new products.
(I work for Google, on cloud, on the containerized developer workflow)
For any multi instance or multi datacenter managed distributed service, lock in is to be expected - the companies build and provide the service tightly coupled with their infrastructure and tech, so the effectively the companies are the service. The lock in follows from there. See DynamoDB, Bigtable, Spanner, SQS, S3 etc.
With s3, at least, Google Cloud Storage can be used as a drop in replacement in most scenarios. The object model is the same, and it even supports the same API.
