Meteor.com free hosting ends March 25, 2016

We are exploring easy ways to keeping the *.meteor.com for specific package demo apps (see that thread) but cannot deliver this wholesale to all *.meteor.com apps for a variety of liability reasons.

4 Likes

Hey great news - we’re going to talk about this on the next TRANSMISSION show, with the Meteor and Galaxy project managers, @zoltan and @rohit2b. Show is recording on Friday and will come out next week! Get notified when the show comes out, and help ask questions here:

2 Likes

Comes out next week on Friday? :slight_smile: Same day *.meteor.com goes down? :slight_smile:

2 Likes

I think a safe assumption to make is that the answer is “yes”. I don’t imagine thousands of dormant apps to be prohibitively expensive. I’m guessing(!) it’s the apps that are actually running that are responsible for most of the (engineering?) cost.

Sure, it would’ve been preferable if this announcement would have been made three months ago. I’m sure MDG considered negative feedback on the short notice, so I imagine the decision was probably made under pressure. I don’t even want to speculate about the cause.

Whatever the reason, I’m sure it was necessary, and ultimately good for the community. Nobody benefits when MDG is in trouble. In the absence of information we could have a bit more faith in what MDG is doing, certainly considering the time and effort they put in their community.

3 Likes

does this mean existing apps will just stop working?

or that they are just not accepting to host any new apps?
i’m really too busy to deal with moving a bunch of apps right now. maybe that’s why its such a short deadline, to try and move people in a hurry to paid…?

Yes, all your *.meteor.com will cease to work. It’s been a pretty bad run for the fortnight leading up to this anyway with timeouts and all.

Just sucks that @debergalis came out to explain that they are in a transition, to be patient as they do the migration to better platforms and then MDG hit us with a 2 weeks notice… bit of a joke how this whole thing is unfolding.

This was over on the *.meteor.com is completely down topic for the fortnight leading up to the announcement. Made me think they were fixing the issue…

3 Likes

I expect the silent majority to show up in a week from now when their projects stop working, cause they’re the very last who will get this information. Then this topic can beat the records. :slight_smile:

10 Likes

What exactly is “this” in this particular instance? Redirects? Why would providing redirects for existing (or even future) projects be a liability?

“This” meaning offering a redirect for all apps currently on *.meteor.com free hosting. While this sounds like a simple fix, it actually is a major liability risk since MDG would have no control over the acceptable use of the destination URL or site. While we don’t expect members of our developer community to be bad actors, with nearly 50,000 apps being hosted on Meteor.com today, it is impossible for MDG to audit or prevent abuse if a MDG-sanctioned redirect was offered to every app. In addition, this would require either a lot of manual work to process redirect requests and/or R&D investment in setting up a self-service tool for process self-service requests and ongoing redirect support.

1 Like

Hm, how does bit.ly address this high liability risk?

And how that it was a problem until now to not just link to them, but also host them?

Could you then just redirect all existing *.meteor.com domains to http://bit.ly/meteor-app-* and then leave people to register their bit.ly redirect and this is it?

7 Likes

We don’t host any apps on *.meteor.com. But as the CEO of a company building an enterprise class SaaS app using Meteor we find it invaluable to see example demos of Meteor package addons that we may want to use. In fact, just today we we found a couple of new packages that we might use and they both showed how they worked on meteor.com. You really should provide some way to redirect these app sites otherwise this transition is going to be even more of a cluster@#$%.

Please let us know about these packages that have good demos on this thread. Keep in mind, there are also options (e.g. Github page) that are likely more useful to showcase actual bits more than a free hosting option.

Well, first things first: we didn’t handle this announcement well. We should have done a better job explaining the reasons behind the decision and helping answer your questions once the post went up. There was a fair bit of one-hand-not-talking-to-the-other behind the scenes on this one. That’s on me and I’ve started a process here at MDG to make sure we don’t flub a communication like this again.

There are a lot of great ideas on this thread. We’ve been reading and discussing them all week. I want to thank everyone who took the time to explain your perspective and help us understand the best way we can help you.

On to some specific questions.

Why the fast 2-week shutdown window?

The reality is the legacy free hosting tier was burying us and we couldn’t afford to keep it up for much longer while also working on Galaxy.

The database was the hardest part, as we’ve discussed here, but there’s more. That system was just never written to support the 50K+ apps running on it. It’s simply not built for what it had become: it has none of the isolation, security, monitoring, and scaling investments we’ve made in Galaxy. It made for an unpleasant situation: we found ourselves reacting to problems instead of being able to proactively head many of them off, and we couldn’t easily put standard processes in place to help us address these kinds of incidents when they did happen. That’s … a bad way to fly.

Time is our most limited resource, but cash matters too. With the growth in apps, the infrastructure takes more and more of it. Fun fact: the weekly hosting bill for the servers that run the free meteor deploy tier? Now north of $10k. A week.

What about free hosting on Galaxy?

We made a promise (with the best of intentions) that we can’t yet deliver on with Galaxy; I regret not being more upfront about this last week.

Free hosting forever, based on Galaxy, had long been our intention. It’s simple to do in concept. But we’ve learned there’s a significant amount of work in building the free tier on top of a production system where we need to ensure high availability and predictable performance for our paying users. We’re not closing the door on a free tier – there are several options we have looked at – but it’s not a quick project and as I explained, we can’t keep running the legacy systems while we develop that capability.

Moreover, the work on a free tier competes directly with other features that customers have been asking us for, such as support for multiple container sizes and for AWS regions in Europe and Asia. It’s a long list. We are eager to build these, and we have to make the best decision we can about how to prioritize them against each other and against a free tier. One guide in our thinking is asking what’s most important for professional developers. It’s the theme of the upcoming Meteor 1.3, and as I said in my Devshop talk last month, you’re going to see us focusing more and more on the features that professional developers most need, across all our projects.

All that said, I think it’s valuable to offer as much of the “one-click” meteor deploy experience as possible. I’d love to find a way to get there, possibly working with a database partner.

Community websites and package demos?

We love the idea of a special program for Meteor-related sites, for example. @marktrang and @rohit2b are working on a sandbox program for that. Expect more from them on this one shortly.

Next

I’ll hang out here on the thread and be on another Transmission soon.

44 Likes

Is it possible to provide a galaxy free hosting where the app will be destroyed in 2 days if there is no payment made.

This helps in the following ways

  • newbie developers can be taught in meteor during training programs with galaxy as part of the curriculum. This will close the development loop instead of introducing them to heroku / bluemix.
  • once someone understand the seamless and easy deployment in Galaxy with his MVP, the probability of conversion from Free to PAID will be high. I feel MDG’s revenue will only increase in this freemium model.
  • 2 days will not be a huge cost for MDG & it creates a sense of urgency for a newbie developer to act on it.
1 Like

It’s this simple guys, really. They didn’t expect such a huge uptake in usage. Speaks volumes on how kickass Meteor is. :smiley:

3 Likes

I wonder if it would have been possible and better to seek more VC funding in order to maintain free hosting. VCs are more likely to provide funding wherever there is great traction. Think of the history of YouTube and maybe many other big names.

YouTube other big names are B2C companies with tremendous amounts of user traction - millions of users. You won’t wow VCs with likely profitability of Galaxy. Userbase is “low” compared to B2C markets, thus scale has to be considered more pessimistically especially since Meteor & Galaxy haven’t proven yet that they can scale this. All they’ve proven is that they can throw an open source framework out for free and people will use it and they have a few paying customers. Technically they’ve just validated their (commercial) MVP in the last months and are ssentially entering growth stage right now so first they’ll have to prove that their MVP can grow bigger and make revenue now.

3 Likes

Thanks for the clarification @debergalis. It seems like some sound reasoning went into this decision and it’s good to hear it. One question:

If one of the major issues is with the number of apps running, is there any way that you could just delete apps older than say 24 hours, so that people have the ability to test out an app (especially a mobile app), for free for a short window of time? As long as people are warned about this limited time, it shouldn’t be a problem and it should reduce the number of apps and the storage needs on your end. The ease of meteor deploy is huge for beginners.

Yes, a free-for-24-hours feature is possible. (I’m actually quite fond of
the idea.) But it has to be a Galaxy feature. We’re not going to run two
parallel deployment systems, especially when one of them doesn’t have the
monitoring hooks necessary to diagnose problems and automatically page
engineers.

13 Likes

As @debergalis mentioned above, we are working with a select group of package authors to migrate their demo/doc sites from Meteor.com free hosting to Galaxy, free of charge. See the details here.

Also, we’ve formalized the ability to request access to Galaxy Community Sandboxes, a free version of Galaxy that lasts for 7-days, no credit card required. While we’ve supported community events (e.g. hackathons, workshops, and Meetups) with generous Galaxy credits in the past, Galaxy Community Sandboxes will make it even easier for event organizers to provide a great deployment option for Meteor apps. We can’t wait to see how the community will put these to use in future events!

16 Likes