Quite cheap and good. I was wondering if there is a features page. Or features comparison with the alternatives. The alternatives have features pages like:
Will the future version of Galaxy (with unit based pricing), have equal support among developer and non-developer. I don’t have a large enough app to have it be worth it to pay $500 when there are cheaper options out there, but it is definitely an important production app to me and the people that use it. Would it be recommended to wait until the unit based pricing to switch or is the downtime issue just a safeguard that will likely not be an issue?
An integrated database option is coming. We just don’t want to block releasing Galaxy Developer on having it – it’s easy to grab a database from companies like https://www.compose.io/ and https://mongolab.com/.
I recently ran into IBM Bluemix and saw couple videos on YouTube about deploying meteor apps there. https://m.youtube.com/watch?v=dyVmB8vHFaE So I’m comparing MDG new Developers Edition to Bluemix. Does anyone have an opinion or have used containers on Bluemix?
Agree with you. It is actually for beta testing. We can register an AWS or other platforms for beta testing as well. For others who have own servers they can install Meteor for beta testing.
The problem is not about spending 13$ on startup hosting, the problem is that this hosting is advertised as not being fit for production. Production for 10 beta testers is still production. I don’t want to pay 13$ per month for my users to think that my app is not reliable because too often down.
Actually i’m using a dedicated server that cost me 20€ each month for an intel Avoton with 8gb ram and a 120 GB SSD. For my beta testers, i find it makes more sense that those 512 MB containers not fit for production. Perhaps there is something i don’t get about cloud hosting and i’m completely wrong.
Could you share your provider?
It’s a french service:
The ISP behind it is Free/Iliad https://en.wikipedia.org/wiki/Iliad_SA, a disruptor in telco in France.
Yeah I get your point, and I do agree with you. I think MDG has a marketing problem. Galaxy is unjustified at $50monthly per 1GB container if I can accomplish the same thing with a Developer Plan at $13 per container.
The way they try to justify this is to remind us that it’s a smaller container, which may restart randomly, and to deny critical support response times.
That being said, if you run on multiple containers it may certainly be good enough for production. They aren’t exactly saying that it will be uselessly unreliable and unstable, just that the Team, Business and Enterprise plans are recommended for production.
Also just because Modulus.io doesn’t warn its customers, that doesn’t mean it’s a good choice for mission-critical production apps. After hosting at Modulus for a year I no longer have any application on Modulus nor would I recommend them for anything very seriously. We suffered availability issue after availability issue. Our apps would spontaneous flatline and we were unable to restart them. Their entire control panel would go offline. They had load balancer issues which lasted weeks at a time.
Here’s an example of the kind of “support” you get from Modulus. I could hardly make sense of it myself, and I only got this message after writing them for a few days of shitty performance. I had my dns configured correctly according to their docs, but they don’t maintain their docs and they don’t proactively tell you when you need to change your DNS to get better performance.
Your sluggishness is very likely due to our Joyent balancers being overloaded. Your application is hosted on AWS, but the *.mod.bz and *.onmodulus.net domains are geo-dns and can resolve to Joyent balancers even though your app instances are on AWS. As a test, if you have a custom domain available, please point it directly to an AWS balancer to see if that resolves your issues. I’d recommend the 54.173.116.166 balancer, it is our newest and biggest. We are replacing all Joyent balancers with much larger instances to meet the demand, but the process is more invasive than AWS, which is why it takes a little longer to scale those up.
It is a high-priority task to better document custom domains and more clearly define how our routing works so customers are able to make more informed decision on their DNS setups.
I thought paying a company to host my app was suppose to make these kinds of problems go away.
So even though MDG warns us about using Developer Plan for production, hosting on a couple spot instances will might be better, and cheaper, than hosting on Modulus.io.
Being located in San Fran, maybe MDG just has higher than usual expectations regarding the definition of reliable production hosting – considering on average eight unicorns were born per year in the Bay Area over the past decade.
So I understand how they are in a difficult position. For example we pay $500 monthly for a 10 x 1GB container Team Plan, but the RAM utilization for all my apps is between 13-30%, even over a busy Black Friday and Cyber Monday sale (we are B2C). So our 1GB of RAM appears to be overkill. At half the RAM i’d only be up to around 60% utilization. Unfortunately they don’t offer less RAM at 1/2 the cost at this service level.
Perhaps i could downgrade to developer plan for some of my apps. If they are mission critical that’s probably not a good idea, but for non-mission critical apps, like our company’s job site it’s probably fine.
If you are self hosting for 20€ monthly I guess there is no reason to switch to developer plan. In my case I’m more of a product manager than a server professional. I’m not interested in learning how to do tax accounting myself or how to host my own applications. I’m happy to outsource the technical grind to others if it frees me up to focus on the product and the market.
It’s a rational decision. I could use my time to build a company that might make me millions. When I add up the hours I spent managing my own DIY server, or the hours I spend analyzing poor performance and going back-and-forth with tech support at some shitty hosting company, I save a lot of time by paying for Galaxy. It all just depends on how much you value your time. Granted $500 per month is a lot of money and I’d really like to see some options between $13 and $500. Bluemix looks promising!
Do you have tried Digital Ocean? I have just attended its seminar. Its all SSD storage looks suitable for web app. You may ask them to pre-install Meteor for you. But private server is still less expensive solution especially for beta testing. For small amount users (less than 1 million) current public cloud platforms do not have main benefits. And when you have heavy database usage the expense may be skyrocket in public cloud.
Digital Ocean is way more complicated to maintain at a production level. At a hobby/tinkering level it’s fairly easy. Most of the time things go right. However, if a bad patch update takes down your server, or your dockerized cluster isn’t working things go south quickly. Once you get past a single instance on Digital Ocean you really have to spend time to get it right.
Exactly! $40-$50 for a high performance 1GB instance or 2 - 512mb instances would be great!
What is DO? Sorry for the silly question, but I’m currently evaluating hosting options.
what? LOL
Do spot instances all fluctuate with a global price or a local price? If the former why wouldn’t all spot instances below a bid price get shut down? So if one instances gets shutdown because it’s price bid is too low, why not all of them? Presumably they are all bidding at the same price.
Thanks, I still don’t see why if one gets shutdown the rest wouldn’t since all spot instances via Galaxy are all presumably the same size and bid price. Why would running multiple hedge against a service interruption if they are all bidding at the same price?
Hi Matt,
Lots of confusion, concern and speculation about Galaxy hosting here. Why doesn’t MDG join in the discussion a bit more and try and address these concerns?
It seems that using a combination of some on-demand instances combined with a diversified fleet of spot instances
http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/spot-fleet.html
Would provide a great price and improve availability and stability.
Hey guys,
maybe some of you can explain me the big advantages to use Galaxy / Digital Ocean / AWS or other virtual cloud providers. I’m just using for several years dedicated servers, or for some smaller projects, vServers. To give you a typical pricing example, I’m using at the moment a Hetzner CX60 vServer for 60 Euro / month, getting 8 vCores, 32GB Ram and a 600 GB SSD, including 30 TB traffic (each TB above costs 1,39€).
If I compare this setting with providers like DO, I’m just wondering why I should pay such high prices for a virtual instance? Also their traffic prices are much higher. The only advantage I’m seeing is the possibility, to buy and start new instances within a few seconds to scale on high traffic impacts.
Hi XTA,
Here are my reasons:
-
Opportunity Cost: - I don’t want for myself or my developers to spend time learning how to host on dedicated servers.Paying for hosting, like paying for accounting services, cloud storage, etc. frees up many hours of our time each year to focus on our business.
-
Service, Reliability - I’ve heard it said that in a restaurant “You shouldn’t know when you’re getting good service.” Meaning a good waitress is invisible. The same can be said of hosting. When it’s done right, you don’t have to think about it. DIY hosting means you gotta Fix It Yourself.
-
Scaleability - I used to host my own ecommerce platform (Magento on a dedicated Rackspace server). Several times our products went uber-viral on social networks. We got so much traffic once, our server just choked for 9 hours straight. I had to change .htaccess to redirect customers to an eBay page because of resource limitations (mostly database connections I think). We got a few hundred sales on ebay, but it could have been much bigger if my server hadn’t cratered during that peak. I moved the shop to Shopify and now it’s on an elastic, auto-scaling platform. Some Shopify stores get featured on TV shows like Good Morning America and do millions in sales in a few hours. if you app gets blogged or tweeted by some meta-influencer, will your dedicated server be able to handle it?
On Galaxy (Team Plan) we have 4 apps running on 10 containers. During our recent Black Friday/Cyber Monday sales we reallocated our containers so our critical eCommerce app ran on 6 instead of it’s usual 3. It’s nice to be able to dynamically reallocate resources with a simple click.
What could be more simple?