How did lambda manage to create this false idea that it makes everything instantly easy? If anything in many ways it makes things much more complex and harder and more expensive.
Serverless/Lambda is a deploy option, not an alternative to writing code. Take Laravel, for example - a popular framework that traditionally is deployed to servers, virtual or otherwise. Well not long ago, they released Laravel Vapor (https://vapor.laravel.com/), a deployment platform for your Laravel code. It takes some configuration, but you can generally take your Laravel app and run it on Lambda using Vapor super easily.
So, while serverless is not a silver bullet, I don’t know why you got downvoted, a combination of using serverless plus managed services can significantly lower the dev and ops overhead of a modern SaaS.
Edit: Dan has edited his blog post to address this comment ... good luck dan with Nodewood! See comments below. We should support a fellow founder and HN community member who is here and listening and working.
Original comment:
It’s not clear until the end of the long post that this guy is selling and all in one development package.
No problem with that but it feels disingenuous to make building a SAAS super hard THEN say “he he I have the solution for you!”.
He’d have been better to say up front his product solves the development complexity problem which looks like this....
Then at the end you’d say “gee this guy is right”, instead of “oh I’ve been played”.
Just did. Thanks for the advice! I spent so much time putting together the list of code that when it came time to plug Nodewood, I didn't have much mental oomph left, so I just put it at the end. Your suggestion is much better.
One should be aware of all the things you'll need to build but balance that with the amount of money available. It's probably still possible to create a sass with just plain php and ftp uploads and that might enable you to get a product out of the door fast enough so that money doesn't run out.
If you have the time and the money, sure, do the best solution.
Special if you're working on a very small scale but plan to hire sometime in the future, remember every little piece of tech adds to the probability the person you'll hire is not familiar with it, so more training will be involved.
I wonder what is the shortest path to deploying a web application that is fully codable (as opposed to no-code), that includes user database/management/auth.
I suspect the closest is the old school libraries like Ruby on Rails and the equivalent in other languages.
My concern with SaaS's are the legal problems, need some blanket ToS I can use, do I need an LLC? Etc... Like this rendering bug caused a seizure give me money, sort of thing. Outlandish but yeah
> My concern with SaaS's are the legal problems, need some blanket ToS I can use, do I need an LLC? Etc... Like this rendering bug caused a seizure give me money, sort of thing. Outlandish but yeah
In the US, anybody can sue you for anything. And if they do, you will have to defend yourself in court.
No, you don't need a LLC, and it doesn't shield you from a lawsuit unless you manage the LLC like a real corporation.
I see this type of comment on HN quite frequently. If you're that worried about lawsuits, maybe starting a company isn't right for you.
Yeah do it or don't easy to talk, I'll probably just find some off-the-shelf ToS/Privacy Policies to go with initially and if it happens to be profitable worry about lawyers later when there is actually money to use.
>Getting all these various tools set up to work together will take some time as well. Just searching "configuring webpack for X" reveals a minefield of blog posts written for various versions of webpack and X. Some will help, some won't, and sometimes only experimentation will reveal which is which.
Yes, I too hate Webpack but it keeps getting pushed on me by companies where they have set up Webpack because it is what everybody uses.
If you're evaluating tools for building a SaaS app, you may also consider using Laravel Spark. I recently used it to build an MVP for a client and was able to wrap up the project in less than a month. The tool chain is pretty robust, and the framework and ecosystem are both fairly mature.
Spark is absolutely one of my inspirations (and competitors). I figure there's enough room in this market for a few solutions to this problem, and there will likely be people who don't want to use PHP (for a variety of reasons) that there'll be space for both of us.
> Some are going to be unavoidable, such as JavaScript if you have a web app with any serious browser interactions
With the rise of things like Phoenix LiveView, Blazor for c#, Turbolinks and Stimulus Reflex for Rails, JS frameworks like React aren't unavoidable anymore.
I'm exhausted before I've even got started. And to think all Zuck needed for world domiation was a P4 with 256Mb RAM and procedural PHP. There's a lesson somewhere.
The P4 with 256Mb RAM I was referring to is the likely spec of the server Zuck was renting for $85/month when Facebook launched. Amazing, considering the spec we have available today for only a few $ per month.
For anyone actually serious about prototyping a SaaS, getting MVP out the door, try firebase. It solves a lot of the problems in the article for you. It’s flawed and annoying but I believe indie hackers still runs on it.
It’s relatively expensive, but if you consider productivity gains it’s basically free for a long time.
This is absurd.
I've worked on very successful SaaS (B2B, not B2C) that dominate their industry, it was started by a bunch of students writing shit in PHP, then got rewritten in Java at some point. That's it. Along the lines it picked more tools and techniques of course, but you don't have to do all that in the first three months.
34 comments
[ 3.4 ms ] story [ 82.4 ms ] threadHow did lambda manage to create this false idea that it makes everything instantly easy? If anything in many ways it makes things much more complex and harder and more expensive.
I've had friends propose serverless startups by comparing $5 a month in hosting to the cents of serverless.
But you still gotta write that app.
Original comment:
It’s not clear until the end of the long post that this guy is selling and all in one development package.
No problem with that but it feels disingenuous to make building a SAAS super hard THEN say “he he I have the solution for you!”.
He’d have been better to say up front his product solves the development complexity problem which looks like this....
Then at the end you’d say “gee this guy is right”, instead of “oh I’ve been played”.
If your making adjustments then I applaud you.
Most of these don't fit into SaaS MVP at all...
Special if you're working on a very small scale but plan to hire sometime in the future, remember every little piece of tech adds to the probability the person you'll hire is not familiar with it, so more training will be involved.
I suspect the closest is the old school libraries like Ruby on Rails and the equivalent in other languages.
In the US, anybody can sue you for anything. And if they do, you will have to defend yourself in court.
No, you don't need a LLC, and it doesn't shield you from a lawsuit unless you manage the LLC like a real corporation.
I see this type of comment on HN quite frequently. If you're that worried about lawsuits, maybe starting a company isn't right for you.
edit: me focusing on the negatives
Yes, I too hate Webpack but it keeps getting pushed on me by companies where they have set up Webpack because it is what everybody uses.
* postgreSQL
* postgREST
* React/TypeScript (with create-react-app)
Hosted on AWS for roughly $50/month (or for free with the Free Tier).
For Rails, there’s https://bullettrain.co/. For Django, there’s https://www.saaspegasus.com/. (I haven’t tried any of them.)
With the rise of things like Phoenix LiveView, Blazor for c#, Turbolinks and Stimulus Reflex for Rails, JS frameworks like React aren't unavoidable anymore.
And I couldn't be happier about that.
My understanding was that he was very focused on reliability since some of the other social media networks were so flaky at the time, like Friendster.
It’s relatively expensive, but if you consider productivity gains it’s basically free for a long time.
On side notes, is there any plan to support React components as well?