Jsonbox – A HTTP based JSON storage | Hacker News Reader

Jsonbox – A HTTP based JSON storage

github.com

104 pointsserious-sam36 comments

Whenever you provide something free, "for prototyping" or "for personal projects", there will always be some moron/asshole who will abuse it by deploying it in a mobile app or something. It happens from time to time with OSM map tile servers operated by OpenStreetMap Foundation. And with Overpass API. This all is for software and data that can be entirely self-hosted, though servers are obviously not free.

They will abuse it, and then when it doesn’t perform the way they want it to they will email you demanding support.

Even though people are questioning the purpose of this project, I think it could be useful for those who are prototyping things and don't want to tie themselves down to a specific database like MongoDB... not that there's anything wrong with Mongo. My favorite feature about Jsonbox is the querying by fields in the JSON data.

I've been working on a similar project, although not optimized specifically for JSON, but rather a generic key-value store with a focus on server-side scripting: https://kvdb.io

It seems like a lot of these storage as a service offerings have been popping up over the past few years because there's something about Mongo/Dynamo/etc. that could be simplified even further. I don't quite know what it is, but it feels like developer ergonomics (i.e., API) and ease of deployment (i.e., nothing to manage) are the big ones.

Agree with you 100%. We were working on things where having some basic KV storage was the easiest. To make life a little easier we put this behind a Map like interface. We can then plug the persistence layer behind it for various use cases.

https://github.com/matdehaast/kv-storage

This surely wont last long, the more people that use this, the more likely that this service gets shutdown due to an increase in cost.

A service like this wouldn't cost that much to run even at a scale that serves tens of millions of requests a day. It's pretty much ideal for horizontal scaling. It's certainly in the region where a benevolent Bay Area dev could afford to run it as a hobby.

Looks like the repository is the server implementation. If you can really use this for critical code, it's probably nicest to set up your own server :)

A single instance of MongoDB and node can handle quite a lot. Anyone who needs this service as a critical component of their product can cheaply host their own version of this or risk downtime/slowness that'd affect their bottom line. (Although for commercial uses you'd use something more featured and secure anyway.)

Neat ! If you are looking for ways to monetize the idea, take a look at what dweet [1] did: you pay to get exclusive R/W rights over a "box".

[1] https://dweet.io/locks

Interesting model:

> Dweets are stored for up to 1 month if they are locked. You can query a maxiumum 1 day of dweet history per request or as granular as 1 hour. Locks cost $1.99/month.

Otherwise:

> dweet.io only holds on to the last 5 dweets over a 24 hour period. If the thing hasn't dweeted in the last 24 hours, its history will be removed.

I’m sure AWS has a REST API to store JSON, but the bar to entry is higher.

Making this free is brilliant.

(1) maybe OP has AWS or GCP credits (very easy to get from these companies’ startup ambassadors)

(2) If there’s ever a greedy actor taking up resources, it seems like they would be relying on the system to the point where they are willing to pay for it.

This is exactly the right way to start a business... Do one thing well in a way that doesn’t scale — this gives you the opportunity to learn from customers and build better offerings.

Why do you think AWS and GCP give out free credits to begin with?

anyone out there studying programming, this is a great type of project to challenge yourself to build.

The codebase doesn't need a lot of dependencies and you build up a foundation you can leverage for multiple projects.

Maybe I have not read the doc carefully enough. But what if you PUT/POST to another existing box_id? Or can other people read from your box if they guess the box_id?

For anyone questioning this project, simpy seems to be a pastebin alternative but for json. Could be useful.

There is a simple Python wrapper if you want to use it from your python project https://pypi.org/project/jsonbox/

It works for all Node.js versions ? I don't see the requirements. It seems great for developpement stuff.

There isn't a client side API. It's an HTTP GET for read, which returns an application/json response, and a POST of the same to create, other http ops for delete (DELETE), update (PUT).

I assume you would use https.request() in node or some abstraction that sits on top of that.

Reminds me of Angular (the online JSON storage service that created, and gave it's name to, Angular.js)

Is the idea to fill the gap between s3 and Mongo (and having to host Mongo)?

I don't get it. You're hosting a service that allows users to store JSON for free? Why? Have you thought about what happens to your expenses when potentially hundreds or thousands of people use it?

You start charging for access.

Maybe the author wants to improve our world? It looks like this is for people with modest needs. Thousands of people shouldn't be a problem at all, with the right architecture.

And if we're talking millions of people, surely some of the can be persuaded to donate a small amount? Hosting isn't that expensive if you go outside the enterprise cloud bubble.

What's supposed to happen to his expenses? You assume that the creator is running the service on a host with auto-scaling enabled? If it runs on $5 DO droplet it will simply not be able to handle all requests. Sounds fine to me for a free service.

You say: "Sorry, we need to start charging" and then out of those hypothetical thousands of people some will decide it's cheaper to pay the service than rewrite their code, so you get yourself some customers...

Awesome for small projects. I’d use it. But:

- IANAL but it seems like a GDPR nightmare (and same for porn and legal nightmare),

- It would be jnteresting to see whether some people are careless about privacy and encryption, or use it as a coordination server for viruses or DDoS. It would also be interesting what genuine usages come up from such APIs.

- No-one would use it for a too serious project anyway, because they’d be afraid to lose their data. But if they want, then they can host it themselves. It would just be interesting to have out-of-the-box authentication and password management, but if that’s needed, they’ll build it themselves.

I sure am keen to see some of this JSON porn

{"porn":"\x89PNG\x0d\x0a\x1a\x0a...

Hot AST exports in your area.

How is this superior to <any_other_http_based_storage_engine_with_API>?

It isn't.

And that's fine.

People can build and release their projects without necessarily having to build the best-in-class version of whatever it is. This is a useful, well-written, open source app that someone has taken to the point where they're willing to put it out there for the world to use, and that in itself is well worth discussing and celebrating (as much as a handful of upvotes on HN can be a celebration of something).

To be honest, even if it was much lower quality code that would still be really good. HN is all about startups and the release-early-release-often approach to building. This is exactly the sort of project that deserves to be on the front page, and exactly what I want to be reading about here.

Congratulations to the author.

Can't agree more with your post.

DRY in spades. More peace & power to the author and all that. However, as a putative user of the project, I'd like to know what would I gain by adopting it. Last time I looked, that was part of use case discussions. I'd happily settle for 'No gain, but it's been a great learning experience' - that's not as it was presented, though.

DRY means for "Don't Repeat Yourself". It's fine to repeat other people. In fact, it's a good thing, because redundant systems are very useful.

It does have some json aware functionality, like sort, skip, and limit. Also, two levels of identifiers, where some web storage services only have one.

What is the business model here? Is user data monitized for profit?

Millennials invented CouchDB :)

Reply on news.ycombinator.com