For static data app, what are the main benefits of Meteor?


#1

Reactivity and DDP, it’s all very cool ,but there is an increasingly common use case where the benefits of Meteor seems less obvious to me as a newbie.

Example: In this day and age of widely available Open Data, a typical web app might access a database of 100 million static records, and use, DC.js or D3.js to present various slices of this data in some interesting way (aka a data cube with dimensions). Perhaps show some tables of summary data (aka a report) that you can scroll through or download as a PDF. A typical application might be allowing visitors to explore historic employment data from 1980 - 20015

Do the benefits of Meteor still outweigh various traditional alternatives like a simple server side micro framework like Python Flask or Sinatra? Or is the architecture ultimately geared towards real time interactive data i/o.

If not, what would you say are the main benefits of Meteor when building an app where the data is essentially static.


#2

We’re working on a new data stack called Apollo which is optimized for static data and reactivity. It’s still in early stages, but give it a read:


#3

I think Sashko put it pretty well earlier:


#4

Whoa. This looks really exciting. :slight_smile:


#5

I use Meteor for websites with no data at all. Between the build tool, the super-intuitive API and the package system (I know there are a lot of people who dis’ it, but I love atmosphere), I can put up a static website in a fraction of the time that I could using any other tech stack. (Probably because I’m pretty familiar with Meteor nowadays.)

But, it’s also nice to know that if you ever want to add some data driven parts to the site, you can also have a full accounts system with one line in the terminal (and one in a template), and then you have all the power in the universe for managing that data if you need it.

Yes, it’s wasteful of resources. Yes, the user has to download a big js bundle on first visit. But when I’ve only got two hours instead of ten to get something decent together, Meteor is just too tempting.

The same thing applies (in my opinion) to static data. Meteor’s build tool, API and package system make development so epic-ly fast, it’s hard to go past. Static data can be fetched via method calls without having to incur the server-side overhead of full reactivity (i.e. getting data to the client via pub-sub). (Not that the cpu burn would be too much if the data is static anyway.)

But, as others have said, once Apollo is ready for primetime that’ll be the “go to” for pretty much every sort of data, static or otherwise.


#6

I see where you are coming from re. static websites. But, what about hosting? Here’s an aspect of using Meteor that I can’t wrap my head around just yet, but it’s on my to-do list to tackle. I’ve paid for 10 years of $4/month hosting on your typical CPanel shared hosting service. For static websites, I simply use a static site generator (Jekyll). That’s even easier than Meteor, no? For a static data back end, IMO it’s poor design to use a database server like Postgres or MySQL or MongoDB when all that’s really needed is a file based database. In my case, using SQLite with highly optimized indexes offers incredible performance. I upload my 2 Gigabyte data file to my cheap-ass host, and then run a simple python script to access the data. Some of these off the shelf scripts can turn your simple DB into a OLAP Cube, and it’s blazingly fast. Mind you, I only have 200 users at most, which are academic economists that analyze the data from time to time.

But here’s the rub. The typical “cheap-ass” hosting service doesn’t allow you to run Node.js - doh. So I need to explore using Meteor (which I love) as a client only solution, and then connect to my python backend using *.? Apollo client side looks good for this so far, hoping it will make it easier to interact with say DC.js to build UI that allows end user to slice the data.

There are loads of products and tools competing in the BI / Analytics space. Many of them are expensive, so I roll my own because I’m cheap. Perhaps Amazon QuickSight, when it comes online, will blow away the competition both price and feature wise, and I can stop having to cook my own mini frameworks and focus on delivering data solutions for my clients.

How Apollo will compete in this hyper competitive space will be very interesting… I need to read more, have a lot to learn. But it looks exciting… just going through the github tutorial apps now.


#7

GOOD NEWS. I just checked my El Cheapo host, and it seems I am wrong. You now can run Node.js for $3.95 / month. They must have just upgraded the system recently. (A2 Hosting)

I will try to install Apollo on my account now. Stay tuned…