Hacker News Reader

Hacker News

Comment by pc86 | Hacker News Reader

Parent Full thread

pc86

> Now I only wish I could find a way to move all our infrastructure off AWS

If only there were competing cloud/PaaS services.

View on HN

If only there was a way to buy your own colocation racks and host it yourself.

Not realistic if you want a globally distributed highly reliable architecture.

Like 1/1000 companies actually needs that.

If only they were competitive.

They're not? I've always seen AWS as the most expensive of the cloud providers. e.g., GPU instances on AWS EC2 vs GCP.

In my experience a lot of senior ops people are still hesitant about Google's cloud services having been burned by them in the past. Google seems to have some reputation problems, however true or untrue they may be.

I think a big objection to moving off AWS is a result of enterprises locking in to AWS-specific technologies.

AWS has a lot of neat, domain-specific products and managed services, but it's important to recognize those come with vendor lock-in and that's by design.

If possible, build out your stack to avoid dependence on AWS-only stuff. It's worth it to ask "if I wanted to transition away from AWS in six months, is this addition going to help or hinder that goal?" It's a lot easier to migrate something that uses generic technologies like VM instances, vendor-neutral containerization, and managed databases versus something super specific. And AWS has a lot of super specific stuff, just look at their ever-expanding product list [1].

Once you get away from the specialized domain-specific products, you start to realize that AWS pricing is really nothing special, particularly for VM instances as you've noted.

[1] https://aws.amazon.com/products/

Azure is about 2/3 of the size of AWS based on market share. Almost 70% of the market uses something other than AWS.

What is it that you know that all those other well-paid professionals don't?

Azure artificially inflates its market share by throwing in MS Office revenue into the mix, which is pretty misleading.

https://www.techrepublic.com/article/microsoft-keeps-hiding-...

G Cloud was rumored to have to be #2 by 2023 or they'd shut down, and currently they are trailing quite far behind.

https://news.ycombinator.com/item?id=21815260

So does AWS. They include revenue from AWS Workmail and other SaaS offerings. Too bad MSFT or GOOG make more money than AWS with their SaaS or "Productive" offerings.

This is probably true, but such a laughably small number of people use Workmail that I doubt it has any material difference to AWS's reported numbers, so while they might not reflect the true AWS marketshare sans-Workmail, it's probably still in the same ballpark. The reason people bring it up in regards to Azure is because it does likely artificially inflate Azure's reported numbers a material amount and makes them harder to trust.

I don't know for sure, but I'd guess MS Office makes up most of the revenue for their cloud offerings. Honestly I think their cloud is mostly a service to push MS Office harder, and they seem to be doing quite well for that.

My point is every cloud vendor needs to break up revenue by IaaS, SaaS, and PaaS. Each one looks strong and bad across segments and hence such a breakup will never happen. But saying AWS or MSFT is #1 misses the dominant segment where they are leading, hides where they are struggling, and gives an incorrect impression.

On the flip side, it'd be interesting to know how many AWS instances are primarly VMWare instances hosting Microsoft Server solutions (or just directly hosting Microsoft Windows products)?

https://aws.amazon.com/blogs/compute/learn-why-aws-is-the-be...

Reply on news.ycombinator.com