Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

What I am wondering is if the complexity of AWS is required for 99 percentile. There are a lot of niche services and duplicated ones on AWS and a targeted replacement for the most popular ones would be enough for most.


You only require AWS if you build something to run on AWS. That's the thing. You can easily run it on something else, just build it for that specifically. Now, AWS-style services have become somewhat of an industry standard (e.g. S3 offered by countless operators). But still, I think offering AWS style services is a weakness because you can never become better than the original.

And cloud is only really cost-effective when it comes to startups that have not much cash flow but expect/hope to explode rapidly by going viral. Cloud gives them that kind of infinite scaling and the ability to pay as they go (the uptick in clients will pay for the increased hosting when they do make it).

In Europe this kind of business model is very rare though. We don't just spin stuff up like a weather balloon and hope it floats.


Netflix as the canonical example would beg to differ as would Apple who hosts plenty of its services between AWS, GCP and I believe Azure. I only know first hand about AWS since Apple talked about it during ReInvent.

There are plenty of large private corporations and governments who host on AWS. Maybe they didn’t do it naively?


Netflix mainly runs from caching boxes at ISPs though. Most of their content comes from there. AWS is way too expensive to serve all that content from.

I see a lot of dumb implementations. At work we're picking up all our physical servers and moving them to AWS compute boxes that run 24/7. Purely statically, just because our idiot CIO wants to be a "cloud-driven company" so he can spout the buzzwords. We're spending a lot more money to get the same only on someone else's computers and get none of the actual benefits that cloud can offer.


Yes. But Netflix is by far AWS’s largest customer. The distribution happens on caching boxes. But there is a lot of processing that happens on AWS. You can look at any of the numerous reinvent videos or even Netflix’s own blog.

Just because your company has a brain dead lift and shift implementation (don’t do that), doesn’t mean every company does.

As a former employee of AWS ProServe (Amazon’s internal cloud consulting department - full time direct hire employee) and now and outside consultant, I’ve seen and been involved in a lot of large scale implementations.

I have no love lost for AWS the company (https://news.ycombinator.com/item?id=44607821) , but let’s not pretend it’s a bunch of startups and naive enterprise corporations that don’t know tech.


Joel Spolsky (founder of StackOverflow) talked about this almost 25 years ago.

https://www.joelonsoftware.com/2001/03/23/strategy-letter-iv...

Everyone needs a different 90%.

I’m not trying to wear “I worked at AWS” on my chest like a 22 year old posting on Blind. But my experience working at AWS ProServe before working at 3rd partner consulting company I’ve seen a lot of different implementations even though my focus was on cloud native applications.

These are all of AWS’s named specialties.

https://aws.amazon.com/professional-services/

Large companies aren’t going to go to a smaller provider. Yes I know about “hybrid cloud”. But modt companies don’t want to go that.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: