Serverless Still Runs on Servers | Hacker News Reader

Serverless Still Runs on Servers

lightstep.com

24 pointsjoeblubaugh13 comments

I figured most of us knew this by now? Abstracted server infrastructure would be a more appropriate name, imo.

At this point it seems like mostly pedants and literalists. I wonder what the overlap is between “serverless still runs on servers” folks and the people that beat the “cloud is just someone else’s servers” horse to death for years and years. I totally understand if someone has a hard time with nuance and not taking things literally but that’s rare. In my experience it’s just fueled by condescension and/or this assumption that people are too stupid to breathe.

I've always thought "Serverless still runs on servers" and "The cloud is just someone else's computer" aren't just condescending and pedantic, but also highly dismissive.

Yeah, sure, "serverless" and "cloud" are both buzzwords that reached peak buzz a years ago (serverless being more recent), but they're actually incredibly powerful.

Before cloud services became a thing, if you wanted to run a high capacity service of some sort beyond something a simple shared web host could offer, you had to either run your own data center or pay for colocation services, neither of which were cheap or quick to set up. Scaling up and down to handle peaks during specific times of day simply wasn't a thing. If you were launching something new, you had to take a guess on how much processing power you'd need. If you guessed too low, you potentially lost customers as they got frustrated and left while you worked on scaling up. If you guessed too high, you wasted a lot of money. The ability to quickly scale your infrastructure up and down in minutes or even seconds is HUGE.

The ramifications of serverless are equally large, because now you don't even need to consider your scaling at all, not to mention patching your operating system and other software. All that matters is your code. Throw your static files in S3 and let AWS handle the distribution, rather than having to spin up an EC2 instance and keeping nginx up to date.

Snubbing technology just because you have some pedantic reason to hate the name is just silly and gets in the way of discussion of the merits of the technology.

Absolutely agreed on all counts.

See, I use the "There is no cloud, only another person's computer" comment as a lesson to both my students and less tech-savvy friends - for privacy reasons. It may be an incredibly powerful computer, but the core issue is, it's someone else's. You can't control what they do with it, and you certainly shouldn't trust them to have your best interest in mind. I don't care what buzzword you use. If data is stored remotely, it's not secure because the one storing & hosting it can choose to see it.

Good way to teach kids how to be properly paranoid on the Internet and the Web.

It is also someone else's OS, software and hardware and compiler (Thompson,'84) even if you "own" the specific license or a copy . Physical "ownership" is very arbitrary line to draw i if you are paranoid about this kind of thing , Storage does not equate able to see if you control the encryption keys

It is, and I wish I was capable of having it otherwise. I merely lack the necessary skills and equipment. There's a much lower skill hurdle to setting up your own server than there is to writing your own OS from binary up.

I wouldn't make the "the cloud is just someone else's server" argument here, or to a tech crowd in general. But outside the tech sphere, it's not as well understood as you may expect. I've had a decision maker on a tech project respond "are you sure?" when it was put to them.

I don't feel it's condescending to make sure people know what they are signing up for. I've had the "you can't store credit card numbers" discussion with people, who have said "but we're not storing it on a computer, we're storing it in the cloud" and genuinely didn't see the problem.

The way I justified "serverless" to myself is that it's the same broad class of abstraction as saying that TCP transmits a stream while UDP transmits datagrams.

The question really isn’t about whether serverless runs on servers or not. It’s about whether the workflow still feels like dealing with your own servers or not. In some ways, yes it still feels like that and in other ways, such as scaling up and load balancing, it doesn’t as much.

Obviously...

Are you sure it runs on servers? I thought it runs in the cloud

lol!

Reply on news.ycombinator.com