The idea of the project is great. It can be difficult to get folks up to speed on various Node JS topics, and one of the big ones (like with any code deployed on a web framework, really) is getting people up to speed on deployment. Anything that is working on making that easier, especially for new/young developers, is great.
That being said, I really do not like the pricing strategy because I think it completely incentivizes bad behavior and optimizing your API/apps for weird things at anything past micro-project scale. Things like when database queries happen are also not entirely opaque to developers, especially new developers, which is going to lead to them not understanding charges (for instance, would a new developer understand that on a fairly basic installation of Passport and Express, every page request, logged in or not, would result in DB activity as Passport tries to figure out if the session is logged in or not? That's likely to far overshadow the single API credit needed for the API hit.) I think that Heroku's free tier would be much more attractive for small projects, and for anyone more advanced than that, a $5 Digital Ocean droplet is simple enough to learn very basic server administration and quickly deploy your application. There are a number of other Node platforms (Modulus) who will give you good service for < $15/month.
I think it would make a lot more sense to structure this more like EC2's burstable instances, which accrue CPU credits for times when they do not run at their allotted CPU, and then can expend those credits to burst above that CPU for smaller amounts of time. That achieves a lot of the good things you're working on, but doesn't push people toward minimizing trips to and from the DB or instance.
Thanks for the insight. :) I don't disagree. Database queries don't use credits right now. We're going to be playing around with what makes the most sense for our customers. Expect things to change as we grow!
I've met Keith a few weeks ago at the NodeJS NYC meetup. Great dude, who genuinely wants to make development better. Polybit seems particularly cool for front-end / designers, mobile devs, and anyone else who'd rather build an app than fret over the high availability, scalability, or etc.
What I'm saying is: cool idea + Keith's very approachable if ya wanna pick his brain about the design :)
I want to second this. There is no doubting Keith's enthusiasm and attitude for both development and helping end users. If you have any questions or issues to raise, I'm pretty sure Keith will do his best :-)
The premise of this is a great idea. The sign up is pretty neat - def. suited to expected audience/users. However, no pricing info? I'll come back once pricing is displayed.
I realize this style of pricing is the new hotness and I get why it's appealing for the company providing the service, but as a developer this is really unappealing to me. The last thing I want to do is have to worry about how many API requests my application is making because it costs money.
For example; if I want to render a complicated page out and the API design will be better if that's done with 10 API requests then I want to make it with 10 API requests. I don't want to have to feel like I need to make some kludgy nasty single request in order to reduce my costs by an order of magnitude.
Similar with queries; I'd rather use multiple simple queries than craft some nasty monstrosity query that gets all the data I need at once.
Am I just crazy?
Full disclaimer: I work for a more traditional web host. But these views are totally unrelated to that (I think), it's just based on my personal thoughts/take on this pricing structure.
PS - I really dig the design of the site and think this is really cool and everything. Just the pricing structure seems like a friction point for me personally.
Sometimes it is better to do many smaller requests versus one bigger request, and it would be nice to see a pricing structure based on bandwidth at least.
How often do the credits refresh? Once per day? Per month?
If it's per month it feels very expensive to me especially at the lower tiers compared to the number of transactions I'd expect to be able to do on say EC2 or GC for similar money... Am I just not the target audience?
This pricing level really limits what kind of apps would make financial sense. Nothing ad-supported for a start.
And another example since IoT is mentioned on the website. Imagine I'm selling a temperature sensor gadget that reports once per minute.
That's 2 credits * 60 * 24 * 30 per month. Even at the highest level that means 115 devices for $999 per month. If I want to sell the gadget for $10 one-off (not a low price taking into account retail margins) you can see the problem.
It means you get 2x the credits because they're in beta. So presumably 10,000 (x2) means for the price on the right column, you'd get 20,000 credits. Just a guess.
We're early and we're super grateful for anybody who's willing to test drive the services. I expect I'll be giving away a lot of free use to active early adopters.
Keith is awesome. I've met him a few times to chat about Polybit and he is seriously dedicated to making a great developer product and solving the pain of standing up a backend API.
Thanks :). The truth to anybody reading is that James is the awesome one. Since I first met him early this year to talk about his experience at Parse, he's been one of the most supportive people I've met in Silicon Valley to date.
If I can give a piece of advice to other early founders: it's tough, it never seems to magically get easy, but there are truly awesome people out there willing to help. Still a long road ahead, but I'm very, very grateful to everybody who has helped make Polybit a reality so far (mentors, advisors, and especially the developer community).
Pretty cool stuff. I got stuck a bit in the registration though. I always use a password manager and paste in my passwords. It's telling me it doesn't like my 16 character password because it needs to be 5 or more, and when I paste it, it's in cleartext.
This looks really interesting but I personally really dislike the pricing, maybe others will love it but to me it's nearly impossible to guess how much this will cost me. When I start a new project I have no clue how many requests or queries will be made. I get the advantage of not working about infrastructure but I'd rather pay a flat fee and then be able to scale up to handle more requests.
Great to have this feedback! Nothing's set in stone yet. This is what we're launching with but I'm sure there's a good chance we'll have something more palatable for you down the line, if you're interested! :)
I'm confused on my credit balance after signing up. The pricing printout shows a 1,000(x2) for the registration tier, and below the table seems to indicate that these (x2) markers represent the double credit offer for beta users.
However, the account I just registered shows a balance of only 1,000 credits. Is the credit bonus only applied to credits which were paid for? If so the (x2) in the first row is a bit misleading.
That being said, I'm excited to check out this service and don't mean to complain about getting something for free.
My bad. Misprint. Happy to honor it though, and give you an extra 1,000 credits. This goes to anybody on HN - E-mail me (keith at polybit) with the e-mail you used to sign up and I'll add the credits by the end of the day!
Oh, no need, I figured it was a misprint and just wanted to make sure you didn't get any unexpected complaints about it. Thank you for being so cool about it though!
Yeah, some other users are reporting issues with Firefox. Putting out some other fires at the moment. Hang tight! Thanks so much for your patience and for trying it out (and reporting the errors)!
What would be alternative deployment approach for nodal apps? It is still just node.js app so I should be able to deploy it to phusion passanger or dokku, right?
I think parent was trying to express approval for Polybit's "API First" philosophy - which would result in a situation where Apps call APIs which call Databases.
Their experience as a DBA being what led them to this conclusion.
The idea of the project is great. It can be difficult to get folks up to speed on various Node JS topics, and one of the big ones (like with any code deployed on a web framework, really) is getting people up to speed on deployment. Anything that is working on making that easier, especially for new/young developers, is great.
That being said, I really do not like the pricing strategy because I think it completely incentivizes bad behavior and optimizing your API/apps for weird things at anything past micro-project scale. Things like when database queries happen are also not entirely opaque to developers, especially new developers, which is going to lead to them not understanding charges (for instance, would a new developer understand that on a fairly basic installation of Passport and Express, every page request, logged in or not, would result in DB activity as Passport tries to figure out if the session is logged in or not? That's likely to far overshadow the single API credit needed for the API hit.) I think that Heroku's free tier would be much more attractive for small projects, and for anyone more advanced than that, a $5 Digital Ocean droplet is simple enough to learn very basic server administration and quickly deploy your application. There are a number of other Node platforms (Modulus) who will give you good service for < $15/month.
I think it would make a lot more sense to structure this more like EC2's burstable instances, which accrue CPU credits for times when they do not run at their allotted CPU, and then can expend those credits to burst above that CPU for smaller amounts of time. That achieves a lot of the good things you're working on, but doesn't push people toward minimizing trips to and from the DB or instance.