by duxup on 6/28/23, 4:24 PM
"We raised a bunch of money"
Generally speaking (I'm not a fly.io customer) at this point that phrase WORRIES me as a customer of any service.
It makes me think that the company exists just on a constant flow of VC or similar money and what I'm using or paying for isn't a realistic or future cost and I'm involved with a system or platform that is getting stretched larger and larger regardless of income and the more money they raise the harder the fall will be.
It's unfortunate but for me it's not a phrase I want to see. I want to hear how they are profitable based on their operations and paying customers or other non VC or IPO type income. That gives me confidence in their stability.
To be clear that's not a disagreement with the title or blog, seems like a fine article that is fairly up front, but it's hard for me to really see "we raised a bunch of money" as a positive from a customer perspective anymore.
by geraldwhen on 6/28/23, 2:28 PM
“ The result of this is an Internet where all of the world's CRUD apps are hosted in Loudoun County, VA (motto: "where tradition meets innovation"), at Amazon's us-east-1 in Ashburn, a city with so many Rails apps that one of them was elected to the county Board of Supervisors.”
So true it hurts
by xianshou on 6/28/23, 2:39 PM
Self-aware commentary like this only works if used sparingly, but this one is extremely effective - acknowledges the mercenary motivations, implicit hypocrisy, and self-aggrandization of such announcements while nonetheless communicating all the important information and succinctly breaking down a real problem that their platform solves. On top of that, the legitimate uses of funding are clearly identified, answering any questions about both "why" and "how much".
Cheers, guys.
by fishtoaster on 6/28/23, 2:52 PM
"Why do startups write announcements like these?"
I've been at a bunch of companies with a bunch of raises. 100% of the time, the announcement was an excuse for press. If you can come up with any excuse to get an article published in a bunch of tech press (other than "CEO arrested for embezzlement + harassment at the same time"), you get a bunch of free advertising. Bonus points if your target customer tends to read tech press.
There's nothing wrong with that, to be clear! I'm just surprised they acknowledged mercenary reasons without mentioning what I've seen as the most common one.
by gizmo on 6/28/23, 2:58 PM
Okay so they raised 100m which means they need to work towards a 1bn exit. And if they want to keep their momentum going their r&d cost will go through the roof. Because they have to build the whole stack and it all has to be rock solid (eventually). It’s an immense engineering effort.
I get how this looks profitable on paper, because Fly charges for compute, and hardware and bandwidth are cheap. But the real expense is r&d, marketing and SBC and those will rule out profitability at any scale. These are not one-time expenses. You have to keep spending forever or your customers will leave. We’ve seen this time and time again. Heroku isn’t the only example. It’s the tragedy of platform/infra startups.
by killthebuddha on 6/28/23, 4:44 PM
It's always seemed to me that Fly has a really awesome, modern perspective on software engineering and that they'd be a great place to work, but I've also always felt like there's a glaring inconsistency between their marketing and product. What I mean is it seems like their marketing (and docs) are all about how trivial it is to spin up globally distributed app servers but the only customers who would need globally distributed app servers are almost certainly not worried about spinning up app servers because they've already got extremely mature platforms with globally distributed data stores. It makes me feel like I'm misunderstanding something fundamental about internet application architectures.
by lawrencechen on 6/28/23, 2:44 PM
> There are two kinds of platform companies : the kind where you can get your Python or Rust or Julia code running nicely, and the kind where you find a way to recompile it to Javascript.
Playful jab on Cloudflare and Deno. Would be interesting if Fly took up hosting "WASM containers" in addition to Docker containers.
by stephen on 6/28/23, 3:13 PM
I really like their acknowledgement that they expected to be used for game/edge caching/etc, but in reality everyone is using them as a new Heroku to deploy Rails+PG crud apps.
Afaiu they're asserting their next goal is to make edge-ify-ing Rails+PG apps trivial / a "two hour problem".
That's pretty bold...historically that's required very app/framework/domain-specific concerns/coupling around "what's the best way (if possible) to shard this domain model", and solving it as PaaS/IaaS basically means writing/running a distributed CloudSQL / Aurora.
Which I guess if it's "just" running an Aurora clone/neon seamlessly in their stack, that seems doable/achievable + very worthwhile as a value-add/product.
by styren on 6/28/23, 3:17 PM
I'm really struggling to understand fly.io's path to profitability considering the relatively low margins for SMB/hobby clouds. They could have the whole world on their free tier but what happens when it's time for EQT to cash out?
Can they build enough features to make fly.io a serious option for companies? I just can't see myself using it or pushing for it at any of the companies I've worked for unless it's a <5 person team with no need to scale in sight.
by jacobsenscott on 6/28/23, 3:51 PM
How is fly.io different (or even better) than just fronting your entire app, including the dynamic content with cloudfront or another CDN? The CDN is close to the client giving you low latency ssl setup, which can be significant.
I get that this doesn't push your app servers to the edge, but even if you did that you still pay the latency cost of hitting your database in the bunker in Virginia. For many (most?) apps this seems worse as you are making many round trips to the db for a single web request. I would rather pay the latency cost on the web request, and have fast access to the database from the app server.
Seems like fly's target audience is mostly static content?
by smallerfish on 6/28/23, 3:06 PM
Congrats! Please focus on getting your documentation cleaned up, your cli consistent, and bugs excised. Your platform is good, and doesn't need too many additional features. Make it all cleaner, rinse and repeat.
by yolo3000 on 6/28/23, 2:35 PM
So if the bookstore doesn't think it's worth it to spend more than 2 hours on deploying to the edge, maybe it's not that important to them, or :) ?
by mirchiseth on 6/28/23, 5:06 PM
by h_mirin on 6/28/23, 2:41 PM
I like fly.io because it's just (much) better Heroku.
I also liked Heroku, so it's sad to see their status now.
by 0xbadcafebee on 6/28/23, 5:03 PM
It's gonna be a sad day when they get bought by a larger company and this writing style is crushed by corporate monotony.
by cvwright on 6/28/23, 2:56 PM
> There are fun, technical, “control your own destiny” reasons to rack hardware instead of layering on top of commodity clouds. But it's really just economics. ... Hardware is what makes the margins work.
Amen to that.
Nice to see somebody who has raised a ton of money, and yet is still being smart with it.
But I suppose the difficult question is, "Why shouldn't Fly's potential customers also run their own hardware?"
by gaffneyc on 6/28/23, 3:07 PM
I'm a pretty happy Fly.io customer and glad to see the direction things have been going overall. While it's great to see Fly is getting more resources to continue improving and building the business I worry about the inevitable VC Countdown Clock to Exit.
What does this new round mean for Fly's long term independence?
by joshcanhelp on 6/28/23, 2:48 PM
A but if a side note but … their jobs page is pretty great. Nothing available but they already have very specific and interesting descriptions written for all the positions they have or expect to have. Feels like they are being very mindful about the company they are building.
by buf on 6/28/23, 2:31 PM
Having grown up on Heroku, it's such a breath of fresh air to see companies innovating in this space. Render, Fly are the big ones, but even little one-man shops like Hatchbox are good enough to deploy medium-scale production apps.
Kudos to these companies, the VCs, and the beta testers.
by lobo_tuerto on 6/28/23, 3:53 PM
Hope you can implement a dark mode for your blog/site now that you are big.
Or at least make it friendly to Dark Reader (a well known browser extension for sites that don't implement a dark mode). Currently all your code blocks are black rectangles under it (easy to turn the offending style off using dev tools).
by carlosbaraza on 6/29/23, 6:54 AM
Edge computing is definitely the last way hosting providers have found to vendor lock you in. Many companies like 37Signals realize that all the fancy cloud provider services have a huge cost because they literally hold you hostage to the mercy of their pricing teams. In a time of high interest rates, I deeply value our commitment to open source container solutions. The only thing we can't do at the moment is to spin an edge network; that's their mote, and why keep doubling down on edge. They double down regardless of knowing most apps top performance offenders don't include the latency to the end user. The bookstore they mention in the article would be much better off deploying all containers in a single cheap commoditized VPS with docker compose, that'd reduce latency between services, and then they can concentrate on optimizing their app and their DB queries. Don't waste time in saving a few milliseconds between your server and the end user, opening a massive can of worms that includes distributed databases, complex infrastructure, vendor lock in, and many other issues they'd happily charge you for.
I wrote a simple library to deploy with docker compose that includes logging, metrics, and a few other utils. https://lostdock.com
by syrusakbary on 6/28/23, 3:39 PM
I wish they mentioned how their service competes against new-comers to the space such as Wasmer Edge [1]. (they kind of mention Javascript, I guess in reference to Cloudflare Workers, but they miss Wasm somehow!).
In any case, congrats to their team and VCs on the round!
[1] https://wasmer.io/products/edge
by mattmaroon on 6/28/23, 2:55 PM
One of the best-written press releases I’ve seen for sure.
by da39a3ee on 6/28/23, 3:39 PM
Can someone remind me, what is "edge"? It it hosting server-side applications in a position in the network topology that minimizes latency to end users? Or is it that, but for static resources (so similar to the concept of CDN)?
by whitepoplar on 6/28/23, 3:20 PM
Since you're hosting your own hardware, one request: if you end up doing managed Postgres, please give customers as much fast, local NVMe storage as they can use. 1M IOPS? Done. 10TB database on local NVMe? No problem.
by AndrewKemendo on 6/28/23, 2:59 PM
Will you be publishing a follow up blog post about how you’re increasing API user fees once you’ve monopolized your particular market? Or how about how you’ll be increasing margins for investors in 5 years as you prepare for your IPO?
Why won’t you suffer the fate of every single other tech company that raises a shit load of money which is completely and irrevocably selling out any pretense of being beneficial for customers and employees (primarily) in the extreme long term?
My new heuristic is that I avoid every single company that raises venture funding. Hopefully others adopt this heuristic because by raising tons of money, so you are explicitly creating an adversarial relationship between the customers/users and your investors so everyone but your founding team and investors in the long term is worse off.
Edit: I’ve been a HN power user since 2012 - don’t ask me why I’m here.
by asadm on 6/28/23, 3:01 PM
> There are two kinds of platform companies : the kind where you can get your Python or Rust or Julia code running nicely, and the kind where you find a way to recompile it to Javascript.
Throwing shade at cloudflare edge workers?
by dools on 6/28/23, 3:13 PM
Have you had a crack at making VoIP call quality better globally? It’s one of those things you can’t cache but I bet there are still some fancy things you can do to speed it up.
by theossuary on 6/28/23, 4:49 PM
Is there any interest in working with telcos to manage MECs? It seems like they've finally given up on building their own container orchestration, and are now looking at Google's GDCE & Amazon's Outpost. But those fundamentally aren't good models for MEC.
I could definitely see a really interesting future for Fly where you're building tech to actively follow users between towers to provide the lowest possible latency to apps.
by miahwilde on 6/28/23, 9:41 PM
Here's what we believed in 2023: people want solutions they can understand in two hours.
Here's what everyone actually wanted to talk about: Trust.
Here's what we missed: ...
by h1fra on 6/28/23, 2:36 PM
Congrats!
I love Fly.io, used it for side projects and I was very pleased with the UX and the free tier.
The blog post is funny but the style is a bit excruciating to read.
by Scarbutt on 6/28/23, 2:51 PM
Why is Fly.io so praised here? couldn't run any production workload on them cause of all the technical issues they have.
by sriram_sun on 6/28/23, 6:35 PM
I really like the "Here's what's not changing" section. Honorable mention for Twilio. Also, the kind of platform companies where you sign up and play tend to get acquired and become the kind where you sign up and a sales person calls you!
by robertlagrant on 6/28/23, 6:08 PM
I'm in health tech and one killer feature would be being able to route and store an individual's requests through and data in a particular location. So I can deploy a health app backend that allows its users to decide where to put their data.
by davepeck on 6/28/23, 3:05 PM
My impression is that when fly.io has been discussed in the recent past on HN, the conversation tends to be less about databases and more about Heroku alternatives. (?)
Out of curiosity, how _does_ fly.io see itself in relation to Heroku?
by tl on 6/28/23, 5:10 PM
Minor nit:
> The kind of platform company we want to be hasn’t changed since 2020. Our features are all generally a command or two in flyctl, and they work for any app that can packaged in a container.
can be packaged
by muhammadusman on 6/28/23, 3:35 PM
Having left DigitalOcean for Fly.io the DX is definitely an improvement. I don’t have any huge apps running there so can’t say how it scales but starting out has been super easy.
by catchnear4321 on 6/28/23, 2:55 PM
> Our first reason is obvious, and mercenary.
even though there’s a reason (or three) behind the style of the post, including the use of transparency, it is refreshing.
by examplary_cable on 6/28/23, 2:46 PM
I'm currently using fly.io and I have no complains so far(most of the problems I had were my fault like a DNS/IP/Certification problem).
by yieldcrv on 6/28/23, 2:40 PM
so I'm guessing Vercel and Netlify deploy cloud functions only to the "least worst datacenter"?
while Fly.io deploys compute instances to more regions and also has a database?
asking because I've been content with Vercel and Netlify's CDN for frontend assets and simply stopped doing system design around relational storage, when I want to stay on a free tier.
by kart23 on 6/28/23, 4:23 PM
by kayodelycaon on 6/28/23, 2:34 PM
> What people actually wanted to talk about, though? Databases.
And they completely avoid discussing it again in this article.
It would be nice if I could know what their product is for because it doesn't seem to be the apps I work on.
by dccoolgai on 6/28/23, 4:26 PM
I don't understand all these comments coming out of the woodwork like "oh, no - one day they may raise prices and try to be the bad guys". I don't plan on using fly personally, but I don't understand how you could look at another provider entering the market and parse that as a "bad thing".
by moneywoes on 6/29/23, 2:37 PM
For the uninitiated, what does Fly do that Amazon does not through AWS?
by wesleyyue on 6/28/23, 4:12 PM
Please implement better support for rolling back deployments!
by techn00 on 6/28/23, 2:33 PM
Does fly use Equinox or something similar for their hardware?
by isoprophlex on 6/28/23, 3:24 PM
> And hardware-accelerated inferencing.
:eyes emoji: fly ppl please do this
by GGO on 6/28/23, 3:14 PM
Fly.io might be an underdog now, but they will come out on top in 5~6 years when most of the small to medium size businesses will be running on it.
by pier25 on 6/28/23, 3:35 PM
Congrats Kurt and the Fly team!!!!
by nickstinemates on 6/28/23, 5:56 PM
I am a big fan of the next gen heroku competitors. At the top of the list are fly.io and railway.app.
by consultSKI on 6/29/23, 7:24 AM
Okay, help me here... with Dockers why would I want to go back to VMs?
by OriginalMrPink on 6/28/23, 2:44 PM
The headline and even the article reads like a big nothing burger.
by secondcoming on 6/28/23, 2:47 PM
Wow, it took a really long time to get to what this company claims to do.
Spinning up VMs. It's not difficult to do this on AWS or GCP.
by dorongrinstein on 6/28/23, 8:09 PM
Fly.io marketing is top notch. Unlike fly.io,
https://controlplane.com lets you run at the edge, super easily but on any compute substrate, not just on one vendor's (Equinix). Your own machines, your own AWS, GCP, Azure, Oracle, Hetzner, Linode, Contabo, etc. accounts, or on Control Plane's many locations in all clouds. On Control Plane you instantly define a "Global Virtual Cloud" made up of any location you specify (your own, or ours, or both). You get a GLOBAL endpoint where a customer in NY gets a response from aws-us-east2 for example, and a customer in L.A. gets a response from aws-us-west2 for example. Of course, AWS is just one of many locations you get run compute on. We're about to release a multi-master Postgres offering and unlike Fly.io, we're 100% K8s, but if you don't know a thing about K8s, you're in luck. You don't need to know a thing about Kubernetes. You simply get secrets management, logs, metrics, tracing, alerting, mutual TLS between services, service discovery, software defined VPN between locations, private link support, direct connect support, your unlimited logs, metrics, audit trail all included and military grade fine-grained access control, audit trail and extreme cost optimization - you pay by the millicore of CPU and MB of RAM. The best part is that you can mix-n-match ANY of the services of AWS, GCP and Azure (Big Table, RDS, DynamoDB, Active Directory, S3, etc.) and not deal with IAM (identity & access management) complexity. Never deal with credentials and go from code --> unbreakable TLS endpoint in 30 seconds. If AWS goes down your endpoint is humming along. You get all the control, without the pain. If you are already a K8s expert and have Helm charts, CRDs, controllers and other existing compute artifacts, we support you out of the box. If you don't know cloud-native - that is OK: just check in your code, the platform containerizes it for you and BOOM - you get a TLS endpoint (with your own domain) and you'll never hear things like "the certificate expired, prometheus fell over, vault needs to be upgraded, we need a new version of Istio, etc.". I urge you to see a demo because a talking about it and seeing it are completely different things. Feel free to visit
https://controlplane.com and I promise you will be shocked in a good way.
Doron, CEO of Control Plane Corporation.