He's making a judgement on the value of aws using a use case that makes no sense to use aws, anyway. He could be doing something similar with lightsail on aws at similar cost if he really wanted to, though.
I think the problem here is that a lot of people use AWS when there's no value in using AWS. It's commonly reached for as a first choice when it's often not a good option if you're looking to run lean.
AWS is extremely popular, but probably only efficient for a small percent of companies who have wildly variable traffic patterns.
I bill between $200 and $250 an hour. (If I had a full-time job I'd be salaried around $90-$100/hour.) Being able to pay me less because AWS's tooling makes life a hell of a lot easier once set up makes a lot of sense even for fairly small companies.
I've watched someone spend days learning to configure a performant DB server on AWS due to their poor disk IO performance and spend an enormous amount to get a high ram instance when a simple SSD based server where more IOPS were trivially available would have had it working out of the box.
Everything has a learning cost though, so perhaps that's a somewhat unfair example as having a deep knowledge of data centers and backbone providers doesn't come free, but I can crank an ansible and docker script out and have pretty much all the advantages of AWS but deployed on my own hardware at a fraction of the cost. So I'm not sure it's fair to say that AWS tooling offers anything particularly unique to merits its market position.
And AWS reduces business risk, too, which needs to be understood and respected, too. Your VPS universe is not actually repeatable. It takes one line to roll out a full environment for more than one of my clients. Dev environment? Here ya go, self-serve bootstrapping and management. Prod environment? Infra-as-code guaranteed to be what you pushed to test at the infrastructural level as well as at the deployable level.
This is why we are replacing system administrators with developers and why we are replacing hands-on system creation with cloud stacks: because it's a difference of kind and the fears of higher operational expenses are trivialized by being able to use that endless inventory to replace the expensive part of your operation--the people.
Endless inventory and minimized administration but higher upfront and ongoing server investment is not a benefit when you need to write a website used by under 10k people a day. Nor is it a benefit for internal services moved to the cloud for nonsense reasons. Both of which I've seen done far too frequently. Simple tooling and small virtual machines or singular dedicated servers like are easily enough for these applications.
I can definitely understand the need for a small team working on a very large, high hit count application though. I'm not saying there are no uses for it, there totally are and you make valid points, just that the uses more limited than one might expect by the hype.
And, unlike the still-really-weird-and-ad-hoc VPS world, stuff like DR is a solved problem when-not-if you need it. The vig for AWS is consistently between 20% and 25% and you are able to leverage implicitly all the incredibly useful tooling and systems around you. If 20% to 25% of $26 is going to materially damage your business, you do not have a business.
0 network transparency
abismal IOPS performance
very limited config. options
Horrible uptime (US East has worth uptime as a region than wast majority of quality DCs)
0 Access to people who can really help you (unless you are at several mil. per month spend)
- Networking is usually real bad. SDNs are your friend. Yeah, learn you an iptables and all, but this is the future, we can do better. (DigitalOcean is almost to the point AWS was at, like, eight years ago with EC2 Classic? Something like that.)
- Geographically-centralized but independent systems are hard to come by and so fault tolerance is a Big Problem. AWS loses an availability zone, my stuff keeps rolling. Can't say the same elsewhere.
- Value-add services. The sibling comment's complaining about RDS, but RDS configuration hits probably the 98% case and You Don't Have To Learn It. I'm a little more hesitant about lock-in services like SQS, SNS, etc., but moral equivalents exist elsewhere for the most part, you can use them pretty effectively.