Galaxy vs Bluemix Meteor Hosting Plan Comparison

We launched an important new eCommerce application into production on back in late July. Within a few weeks we started having various problems with availability and performance. Sometimes the app would flatline, and we were unable to restart it without getting help from Modulus customer support. Sometimes the entire Modulus control panel went down. Frustrated with all the time we were spending trying to analyze performance issues (which turned out to be some kind of load-balancer configuration problems, which have been plaguing Modulus for months.)

When MDG announced the Galaxy Early Access program, we were excited to switch. After hearing about the price ($495/month if paid annually) we decided it was too expensive, but agreed to try it anyway, during a money-back trial period, so we could help them test it as give them our feedback. The plan was to go back to Modulus, which was thousands of dollars cheaper per year. But after the trial, but we ended up signing up for a year at $6000 because Galaxy seemed reliable and freed us up to focus on our product development and market growth.

##Galaxy Team Plan

My biggest problem with Galaxy for mission-critical production apps is the price. They just don’t seem to understand the market, offering plans “for personal apps” starting at $13 per month and plans for production apps starting at $650 with nothing in between. I just don’t understand why they are so inflexible. Meteor is very popular with the startup community, and setting the minimum price for hosting a production app on Galaxy at $650 per month seems hostile to startups.

$650 ($495 if paid annually) on Galaxy buys you 10 x 1GB containers. But what if you only need 2 or 3? I just want a one-bedroom apartment, but I have to buy a 10 unit multiplex. At $650 each 1GB container costs $65, but why can’t I buy them à la carte? And what if I only need half the RAM? Even during Cyber Monday our RAM utilization was only around 20%. That would be less than 50% on a 512MB container. Why can’t I pay half as much for half the RAM?

I think part of the reason is that Galaxy probably expects customers to bother support a few hours a month so then need to pad the price to include support costs. But the only thing we’ve had to contact support about was adding domains to our account (because they don’t expose any way for us to do that ourselves.) So we are probably paying a premium for support which we don’t use. Why not adopt per incident support pricing like Microsoft?

I shared all these ideas with Galaxy team members, and warning them that the community was not going to be happy with plans starting at $500

##IBM Bluemix

IBM Bluemix an open-standards, cloud platform for building, running, and managing applications. Also noteworthy: IBM bought Compose the MongoDB development platform.

IBM has been interested in Meteor for years, and they’ve created short article and video on deploying Meteor apps to Bluemix. Best of all the pricing is exactly what I’ve been tell Galaxy they should do.

10 x 1GB containers on Bluemix would cost $477.75, which is about the same as you’d pay for Galaxy’s Team Plan. But the real advantage is the you can mix and match however you like. 10 x 512MB containers costs only $200.55. Or you could just get 2 x 512MB containers for only $24.15 because they give you 365 GB-Hours of memory for free each month.

But what about support?

Support pricing was a little hard to find. You have to sign up and login, then open the calculator under pricing.

###Free Support
There is a a Developers forum for free support, and they have a stackoverflow tag.

###Standard Support

  • Developers forum.
  • Ticketing system with 24-hour responsiveness.
  • Connect with IBM technical staff.
  • Covers IBM Bluemix services and features.
    10% of account charges, with a minimum monthly charge of ¥21,000 JPY. (sorry my account in in Japan. That’s around $175 USD).

OK support looks expensive, but it seem like you could pay for a month, get all your questions answered, then turn it off for months at a time.

It’s too early for me to say if Bluemix is as reliable and performs as well as Galaxy. We are just getting started with it. But if it does we could save thousands of dollars a year by leaving Galaxy.

In summary, if your hosting needs demand more than a sandbox, but less than 10 x 1GB containers, you may want to evaluate IBM Bluemix. If you actually need 10 x 1GB containers, you might want to stick with Galaxy their plan would include critical support without additional cost.

It will be interesting to see if Galaxy catches up with Bluemix, or if Bluemix continues to outpace Meteor in service and a more enlightened approach to pricing.


Does IBM Bluemix have fully integrated/failproof incremental deployment ?
Still if u want Meteor support, u would probably need to sign for MDG subscription too.

Galaxy support covers Galaxy only.

I am actually waiting for an invite from NodeChef for their great service. It is feature packed. Something I don’t see with the other meteor hosting solutions.

In summary, if your hosting needs demand more than a sandbox, but less than 10 x 1GB containers, you may want to evaluate IBM Bluemix. If you actually need 10 x 1GB containers, you might want to stick with Galaxy their plan would include critical support without additional cost.
It will be interesting to see if Galaxy catches up with Bluemix, or if Bluemix continues to outpace Meteor in service and a more enlightened approach to pricing.

When you buy a game console when its early released, you pay big bucks. After a year, you pay less bucks. And the new game console comes out. You will probably miss the hype and thrills when you wait that year. Or you can try to buy a console at a discounter, maybe a second handed one. It doesn’t really matter. Until that day comes, when your favorite game is released, which you pre-ordered, read every possible post about it and watched every video trailer until your eyes became square and green. The box comes in, you press power and you see it start up…

An error message occurs. Game Disc could not be accessed.

Your heart stops. Tweets from your friends are coming by, and you are not able to hashtag. People apping you asking “Hey hows the new game coming” or have been eagerly waiting for your Twitch going online. You have only one response:


You try to breath. Calling customer support. Enter serial numb-- oh wait, you don’t have it from the local store. You call the discounter, who transfers you to Michael. “Yes, thank you for buying our product. We will investigate and follow up as soon as possible. Can I have your e-mail address?”

“wtf wtf wtf f f f f f f”

You are still not playing your game. Then you discover that its not the console, but the game has compatibility issues with certain versions of the console. “Hope, there is hope!” and tears of joy fall from your eyes, and a smile makes a path all across your face. You quickly follow the threads and see that there are remarks as " \o/ " and “Game Vendor ftw” and all consoles just need to use the update button.

“Errr… Update button?”

You read the manual and you discover that the early versions of the console ran on different hardware with a different build which were incompatible with the Update button. You feel where it is going, but you hope that reading further would take away this feeling. Your console is running that hardware with that build.

Twenty minutes later, while you are still not playing the game, Michael calls you with exact the same news. The good news is that the discounter will release a ROM update which you can download and update as soon as possible, and flash so you can play your game. “We can’t give you an ETA yet - thank you for using MeksBox”, Michael friendly replies to your outbursts of fury.

Bright side. You didn’t die. Only your expectations were crushed. You missed the party. Don’t worry, there will be new ones. Your weeks of anticipation are gone, never to be recovered. You will never be excited again.

If you are a fanatic gamer, you would go pro-vendor. If you have screaming kids sucking the life out of you, you buy a mod chip. Expect not-good-money back when you buy in store, and expect the quality of user experience. Expect kids to break controllers and some games are worth to buy the box for and some should be taken from the discount basket next to lollipops.

Or so they say. Haven’t seen any data on this yet.

Do post a comparison for us when you get on-board. More competition is certainly welcome and benefits us all.

Regarding features, Bluemix and Galaxy also support one-line deploy. Bluemix also appears to be built on Docker Cache sounds like it could be dangerous depended on how it’s implemented.

$495 wouldn’t be bad, especially if it saves you from having to hire another employee and opens up time to work on the actual app itself instead of deployment.

Say you’re paying a senior developer $50/hr— if galaxy opens up 13 hours a month for him to work on things other than deployment, galaxy is easily paying for itself (at the $650 tier).

In people’s experience, can they see Galaxy saving them 13 hours a month in time? Of course some months it’ll save more time than others, but on average-- what are people’s thoughts?

But if you don’t need all that, then cheaper is better right?

Galaxy is certainly worth the money if you actually need 10 x 1GB containers. If your app performs well on just a few containers, then Galaxy Team Plan is like using an tractor trailer to deliver a pizza.

It’s overkill for us, even running several apps. We could get by on fewer containers and less RAM, but it’s take-it-or-leave-it. We have dozens of simultaneous users max. Not hundreds or thousands.

I understand, but does using galaxy save you 10-13 hours a month in hosting/deployment issues? My guess is “no” for you-- but maybe yes for devs who aren’t experts in this area of things.

Is that a fair assumption?

Yes, it was the best option at that time for us because we really were spending too much time dealing with hosting problems at Modulus. But do these Galaxy plans appear to be as flexible as and competitively priced as Bluemix?

Well, I think that depends on the total cost of working with each technology. So,

(Cost of Hosting Option) + (Man Hours Needed to Deploy/Manage * hourly wage) = Total Cost

The first part (cost of hosting option) is pretty straight forward. The second part (man hours needed to deploy/manage) is where I’d be curious to hear your thoughts.

To me, the analysis is worthless unless you’re looking at the total cost.

Regarding deployment time, you can see for yourself. We are talking minutes, not hours.

IBM looks like a good option, going to test it out. I’m not sure what issues you have had with Modules I had issues in the beginning but now there really good.

As for MDG I 100% agree here I think they really messed up.

1 Like

You did see the news about Galaxy Developer Edition, right?


Yes I did, but I excluded it here because in the announcement they mention it’s for “personal apps or early-stage projects” which don’t need same scale, high availability, and high-touch support as production apps. You can only “view the past five minutes of metrics” and they say you can deploy apps for “for small groups of users”. So it sounds it’s more of a sandbox than something for production apps.

They now seem to have flip-flopped entirely so I don’t know what to think about it anymore.

Regarding Bluemix:

(from their site)

Language: NodeJS
Memory: 512 MB
Instances: 2
Users: <8000
HTTP Requests: <600/s

Is 512MB really enough for a medium- to large-scale Meteor app with that many users using it??

Probably not, but I guess it depends on the app! But it’s a much cheaper option to start with and you can always add more RAM as needed. :chart_with_upwards_trend:

I have an educational game app running in production. It is a standardized testing app, so it means, at the time of the tests, hundreds of students are flocking in to create concurrent load with both high writes and high reads. I also keep images and attachments on the db (gridfs) so there’s lots going on.

The app on stand still needs around 180MB (lots of default publications), average consumption goes steadily around 200-300MB range during daily usage, and only heavy spikes during the first few minutes of the tests require around 500MB. And yes, I did see higher consumption a few times. But I guess if one were to design especially the publications carefully, 512MB is actually quite sufficient for many “production-grade” apps.

After all, we have horizontal scaling options and one should be able to design an app to steer it into using two 512MB instances instead of one 1024MB instance.

Hi @maxhodges, as an early stage project, I bet you can access Galaxy Developer Edition (request invitation from my referral) , pay-as-you-go basis of $0.035/GB/hour (~$13/month per 512MB container) (limited to 5 x 512MB)
Did you ask MDG ?