Meteor Cloud was built to be the easiest way to deploy Meteor applications. And we think this is true.
Today you can write:
meteor deploy app-name
And you app is pushed to Galaxy. You can monitor your app with Meteor APM.
What do you think is missing from this model that would make it a high value deployment resource for you?
For me, the following would added value:
- Include MongoDB hosting - yes I know you can easily do that on MongoDB Atlas but as a business owner I prefer to have one invoice for my app and not several
For those apps that need scaling it would be great to have (optional) addons for:
- Zero-config Redis Oplog
- A GraphQL container to easily include external data sources
I’ll add to this.
Multi-region deployments, similar to what MongoDB Atlas has.
I am doing this today manually by routing the traffic to different Meteor Cloud deployments with Cloudflare based on the user’s region.
Fully disable TLS 1.0 and 1.1 so we can pass federal Health IT certification.
I already use Meteor Cloud, but am offering some feedback:
I echo zero config Redis Oplog.
MongoDB hosting would be nice (I know there’s a free option, but a more official option would be nice). Currently we use Atlas but you must use a dedicated plan before you could use VPC peering, which isn’t ideal for smaller apps. Meteor Cloud with VPC peering setup by default would be great.
Part of the MongoDB request is also due to issues with Atlas and +srv strings. If we use a +srv string to connect to Atlas, there’s occasions that have a huge amount of lag from the extra DNS step. No idea why. But using an older connection string instead of the +srv string fixes it. I assume this is something that could be addressed if MongoDB is hosted directly by Meteor Cloud.
I think there’s room for some nice features alongside MongoDB hosting for users that have multiple apps using the same database. Especially considering some articles/blogs I recall in the last year were regarding this subject.
My only other request would be performance for the price of Meteor Cloud. I hadn’t tested this myself, but I had been following past threads questioning the container performance in comparison to AWS self hosting. I know Meteor Cloud comes with a price tag, but I recall the differences was quite substantial. No idea if this was ever addressed, but optimal performance for the price point is always a concern for the long term, as updated versions of our software are being released in the coming months and we will be scaling up the users compared to past versions.
As someone who migrated a few clients off Galaxy to custom deployment: it’s all about costs. And I really saw 66% drop in cost while on AWS (at a cost of doing it yourself, of course; it turned even after a month, though).
- I agree, Galaxy does a lot for you, but so is AWS ECS or a Kubernetes cluster. And while I think it’s fine for one or three containers, but as soon as you have tens of them, that’s a huge difference (and you probably can afford to get like 20h of someone who can set it up).
- Support is extremely valuable, but maybe it should be a fixed price thing, or even a fully optional add-on.
- Professional plan costs between 36% and 100% of the base container price, and gives you access to the APM and autoscaling. At this price, it’s already cheaper to get a Monti APM Pro that gives you a lot more functionalities. And most platforms can handle autoscaling somehow.
- Push-to-deploy is nice, but you can set it up on GitHub Actions or any other CI easily.
The only thing that I miss(ed) from Galaxy is the option to scale containers based on the number of connections. But on the other hand, being forced to scale based on CPU and RAM reduced costs even further.
Keep in mind we are trying to understand:
- What parts of the developer/deployment experience (if enhanced or improved) makes Meteor Cloud a “must use” for you?
- What else needs to be added to or improved upon in the environment and/or the workflow to make Meteor Cloud hosting a high value offer relative to its pricing?
- What else are “willing-to-pay-to-take-the-pain-away” add-on features that if available would make it more likely to use or stick with Meteor Cloud?
Performance is the big one. Galaxy is so expensive for really bad performance. It’s been debated here in many threads so I won’t repeat. It might be an idea for Meteor to clarify what the performance of each one is? ECU hasn’t been used by AWS for a long time but I’m assuming that is the definition Meteor Cloud is referring to? I think there’s a major imbalance between your configurations and what people might expect based on use of EC2 instances. Even a t2.nano outperforms the Compact container by a big margin and I’ve seen people say basic droplets do as well.
Also, on APM, as @radekmie mentions. Meteor APM has stayed stagnant compared to Monti APM, but if you want whitelisting you have to go Pro. Maybe you could do a deal with Monti so at least that’s included ? Or change the pricing for pro?
I would like the ability to not use AWS, but to let another company do the actual hosting.
My view is that we need more professional, (largish) companies instead of just 1 running the whole show.
Cost is important, but I would pay more to be able to switch to non-AWS hosting and get same/bettter performance/scalability etc.
Is this another way of saying multi-cloud support?
The unfortunate part is that Galaxy here is competing with all the other hosting services which normally offer a wide range of services and allow many technologies to deploy.
A must for me is a one click Redis addon to help with scaling (especially if the redis-oplog package is going to be moved into core)
For me some of the following features would be great:
- Apply AWS credits on Galaxy to get discount
- Improve APM and integrate it more into Galaxy (or at least unify the design), as mentioned by others it would be nice to be able to connect the APM to Meteor instances hosted outside of Galaxy as well. I think with the big changes coming an upgrade here will be necessary.
Long term desire:
Additional regions & Multi-cloud deployment, looking towards what MongoDB Atlas does as a golden grail would be a good start. Linode might be a good option as they seem to be cheaper then the other providers.
With this it would be connected a must feature for me which is some sort of a routing system between regions and one deploy to multiple regions.
Agreed with this.
Having Linode, Vultr, Digital Ocean, or OVH as a lower cost option would be great.
If the price is competitive, then the features of the platform become a competitive advantage, and using Galaxy is a no-brainer. Work out a way to make it cheaper.
Ability to include MongoDB is a major plus, as is the Redis oplog.
It’s hard to justify hosting sensitive applications without a SOC 2, Type 2 exam. I have a new app going live this year, and our security team said we can’t use Galaxy b/c of this.
edited to fix typo
User roles! It would be nice to allow users the access to view server logs and container health, without giving them full access to view things like MongoDB connection strings or to make any serious change they want.
Can or can’t? Your message seems to imply “can’t.”
can’t — sorry, typo, will edit original