The thing I miss of the old days was running a single command to publish an app and have it online (on the *.meteor.com domains for example). It was AWESOME!
I think it would be nice (and that many people would want) for Galaxy to be a single-command solution for deploying and scaling Meteor apps including their officially-supported package dependencies (MongoDB f.e.).
meteor create my-app
# write the app
DEPLOY_HOSTNAME=galaxy.meteor.com meteor deploy foo.com
That should literally be it! It would be GREAT! This is part of what made the Meteor experience magical in the beginning, but Galaxy has veered away from that in this regard. This is a valid desire since Meteor apps use Mongo practically by default.
In a large set of default Meteor apps, if there is no Mongo, there is no Meteor.
I voted no. Galaxy folks wouldn’t host Mongo by themselves anyways, they’d use a 3rd party service for that and I wouldn’t even have direct access to the 3rd party’s acount panel (Galaxy folks would have to write their own which won’t provide as much flexibility and data). Also in case of problems or special use cases, I’d have to contact Galaxy folks first, they’d contact the Mongo folks, Mongo folks would have to answer to Galaxy and Galaxy back to me. For twice the price I’d pay directly to Mongo folks.
I voted yes. There is nothing in the poll that states the how. I would be more then happy if they provide me this and have a dashboard available for me where I can log in to see my mongodb stats and logs. Even if they’d use Mongo Atlas under the hood for it.
Exactly, the how doesn’t matter so much. They could in fact add Mongo along with some dashboard for it. The main point is easy-to-deploy apps with one command without having to sign up for other services and make configuration files.
Also, once an app is deployed, the Mongo DB can be accessible online, which means a dashboard for the Mongo can be connected to it as an outside step by the user. So that would also be another way to still have a dashboard even if Galaxy didn’t provide one.
I believe that by default Galaxy can host Mongo, but if there’s really some special case that you can only accomplish by hosting your own, then by all means we should still also have the option to self-host Mongo outside of Galaxy. That is for sure! For example, you would configure your Mongo URL just like you currently do, and it would not be hosted on Galaxy.
Basically I’m just arguing that a default Mongo-on-Galaxy would be nice for most users. When they find they have a special case, they can go to the effort of self-hosting Mongo only at that time, and only if needed.
MDG hosting mongodb complicates stuff in numerous ways, both technically and legally (or so I would assume). I really don’t see the point of having galaxy as a one-stop-shop for everything.
Sure, their old free meteor deployment was cool since it was so easy to very quickly deploy something, but aside small hobby projects it wouldn’t really matter.
You only need to set up mongodb hosting once and through galaxy your deployments stay as simple as before. If you’ve done it before, creating a database with something like compose takes a few minutes tops, that’s the only gain you’d have, a few minutes per unique project and only once during setup. Integrating is literally just creating a user and copy/pasting a connection url.
Then there’s the issue with tailored configurations for your database. How would galaxy price it (would it be an addon?), how would you scale your database, etc. Having those two layers separately (app and database) has a lot of advantages in my opinion.
Also, thinking about the future, I’m sure mongodb isn’t going to stay the only database you can use with meteor by default, so why would they invest into also becoming a database hosting company (especially if they have to support multiple databases in the future) when there already are companies that excel at it.
I’d rather have MDG focus on what they’re good at (improving/expanding meteor and apollo) than dilute their resources from other stuff to create something that wouldn’t solve any real problem.
That being said, if tomorrow MDG announces they’re going to include mongodb in their galaxy plans, I’m not going to disapprove, I’d simply be neutral towards it.
Offering production grade MongoDB hosting is not a small feat. If Galaxy offered it, it would only be a convenience, not an innovative offering, and most likely not worth the trouble of supporting and maintaining such a system.
That is sweet. That’s how easy it should be! (But advanced users should still have options, and the basic easy version doesn’t need to be super robust, just easy, as there’s lots of apps made with Meteor that don’t necessarily need to scale yet, they just need to be up (personal portfolios, experiments, etc)). I mean, I really don’t want to pay for and spin up a MongoDB just for a simple experiment (Meteor makes simple experiments easy to create). Just let me write some code, then throw it online without worrying about extras, I don’t need it to be robust out of the box, and when I’m ready I can find the robust solution for any of my experiments-turned-serious (if any) when the time is right.