Hacker News Reader

Comment by anyfoo | Hacker News Reader

I was an intern at Google right around the time where Google+ (long before it was known under that name) became a thing for Google engineers internally. I sat next to a team that was majorly impacted by that.

It's all subjective of course, but I could practically "feel" Google changing. Before I started my internship and still during it for the most time, Google was this incredible company that people, including me, were in awe about. All their products were so good. So technically refined, so methodical, you could really see that the people working on them were really smart, and really cared about it all.

And then it changed and became this "corporation", and nowadays it's sometimes seen as almost the opposite of what it was, working there being boring but stable. Part of that is probably just a normal consequence of the immense growth, but personally I can't shake the feeling that Google+ had a large part in it.

View on HN

I was a fulltimer there around the same time, and I agree. I ended up quitting fairly soon after the G+ real names fiasco blew up, and it felt like a lot of internal "social capital" got burned on that, which was never really recovered.

I don't think the whole place went to hell at that moment, or anything, and your description is maybe a little more dramatic than I'd go with, as to the role of G+... but it definitely seemed like one of the big turning points.

Sorry, I did not mean to imply a dramatic and sudden change either. I just perceived this as a turning point, like you say (and with only my limited first hand experience there, of course).

Google major revenue stream is and will always be search/ads. Everything else is marginal, irrelevant and that’s why I think they don’t care much.

I still can’t explain how Google got beaten at their own game (cloud / computing at scale) by an e-commerce site : Amazon.

That says it all.

Having many friends who work at Google, I think they have the wrong system of incentives in place. Rightfully, people optimize and prioritize based on their path to promotion rather than building products that people love.

It’s much easier to get a promotion if you build something new vs keep iterating on existing products making it better.

>I still can’t explain how Google got beaten at their own game (cloud / computing at scale) by an e-commerce site : Amazon.

But it wasn’t Google’s game at all: cloud computing is a low margin business (relatively speaking). Amazon knows how to run low margins businesses, and Google doesn’t.

Google was so unimpressed with cloud, that they got serious about it only around 2012, when it became obvious that other clouds will outgrow Google as a whole. I think it was Google’s Internet Explorer moment.

I don't understand how cloud computing is a low margin business.

I can rent a dedicated server at Heztner. 8 hyperthreads, 32 GB RAM, 4 TB HDD. €33.

Amazon EC2. t4g.2xlarge 8 cores, 32 GB RAM, $194. 2 TB EBS HDD $30. $224 total.

Also Amazon charges for traffic.

Surely Hetzner is not working at loss. Surely economies of scale allow AWS to have better deals.

I mean, AWS is literally 7x compared to Hetzner. How is it bad margin?

I'm curious, have you benchmarked both of these and seen if there's no real world difference? I've had major performance differences between clouds on what is supposedly similar hardware. If AWS is faster in nothing then why they're so expensive would be a very good question indeed.

My initial guess would be AWS is relying on its advantage that users who are currently using AWS heavily are unlikely to switch clouds if they suddenly have to go from VM's to dedicated servers, and thus they can profit off such capture. e.g. 7x cost is probably fine when you consider what the cost might be to employ people for a cloud migration.

A lot of it also comes down to reliability, security (not just at the platform level but also the tools for you to build on), and most importantly integration with other services which you’d have to build and maintain on your own.

You don’t buy AWS because it’s cheaper than a dedicated server but because you can get that Linux server using the same management services (API, authentication, monitoring & logging, etc.) that you use to get serverless functions & containers, object storage, managed ML services, normal and reporting databases, development tools, etc.

That’s not saying that either one is wrong, just different - it’s like asking why someone used a hotel for a conference when you had a great tent site on your vacation to the mountains.

The question is how much money Amazon spends on development of those services. What part of price is hardware expenses, what part of price is software expenses.

My gut feeling tells me, that it's not much.

There's definitely a healthy margin but don't underestimate how much time goes into making everything work smoothly. The major cloud providers do a lot of things like handling zero-downtime migration between failed physical hosts, monitoring for infrastructure issues, firmware testing/security/updates, etc. which is completely transparent to you until you realize that you haven't spent time in years on things like hardware or environmental failures (stuff like that SSD 40k hour bug which showed up here, recently, for example) or dealing with stuff like Spectre.

The big confound here is scale: AWS can amortize the cost of someone really digging into firmware security over millions of customers but a smaller provider can't. You might not notice that until you either need to get your environment certified or some kind of vulnerability / supply chain attack against a hardware vendor is revealed. That can mean that AWS is both making a hefty profit reselling those engineers' work to all of their customers but also that it's still cheaper to pay them than it is for you to do that work yourself.

Thanks for that perspective – your subjective impression certainly matches my outside experience. That seemed like the point where the halo faded.

It didn't so much fade as that it got switched off. Suddenly Google was just another evil US corporation when before you might have had the impression that they meant what they said. To be fair to your comment: there were some earlier signs but they could all explained away in one way or another, but that time the mask fell and they never managed to put it back on again.

Now they get a lot of business because they are the least bad option. Which still makes Google useful but I'd love for some actual competition. But with the network effects at play any viable competition can be stifled.

Reply on news.ycombinator.com