This hit a nerve for me. I keep going back and looking at Engine Yard Hoping there will be something there for me but I just can't justify moving from Heroku for the ruby stuff. They have so many knobs to turn and a nice low entry point for very small apps that can be cranked up easily. It's this low barrier to entry that Engine Yard is missing and free for three weeks is not a substitute for this.
AppCloud seems to be (and I hate to use this word) "just" a thin wrapper around EC2. At least I get that impression looking at the price points - the smallest 32bit instance is just shy of 1k/year. Am I going to get more performance out of three Dynos at Heroku for about the same cost? Maybe maybe not. But where am I going to put my staging test/servers? On a 20/month VPS? On an instance @AppCloud that I spin up and down? No, I'll tell you where, I'm going to stick it on a Dyno or two at Heroku and not worry about it. And then when I need to scale or the product is ready for production where is it going to be hosted. That's right - the same place it's been validated and tested on.
By the way, Engine Yard is not the only place that has this barrier to entry problem. Most of the cloud providers do. Joyent.com, Media Temple, Rackspace cloud all have Barriers whether it's poor documentation, higher than normal base costs and bandwidth fees or just inconvenience like having to wait on a call from someone in Texas in order to be allowed to create instances.
If the other players want to compete with Heroku on the service and ease of use side or Amazon on flexibility and scale then they are going to have to do much, much, much better.
Agreed; after using their product for many months, I learned how it is little better than running your own instances. There are some cool things it does, but you end up doing plenty of sysadmin work yourself, writing Chef recipes for anything not supported by their (aging) platform, and then you lose the benefit of support as well. Did I mention that support costs start at $275 (12/5) or $475 (24/7) a month?[1] Without it, you can't open a ticket.
It's also increasingly annoying to see shots taken at Heroku in their blog posts and on Twitter. It's unprofessional. Compete on features and services, please.
"It's also increasingly annoying to see shots taken at Heroku in their blog posts and on Twitter"
Really? We've always been friendly with everyone in the Rails deployment space, and I don't ever remember taking shots at Heroku specifically, though it's certainly possible that we've expressed disagreement on technologies or practices that they and others may use. I'd appreciate it if you'd point out what you've seen, as it's certainly not part of our philosophy or business practice.
With respect to your comment about our "aging" platform, we all age a bit every day. There's tremendous work underway at Engine Yard, some visible, some not. This is no-doubt true of every competitive platform. Many of our customers choose to describe this as maturation! :-)
Yes, all platforms that are static and curated in this way are aging every day, and Heroku has some flaws in this area as well. But as I understand it, the base platform of AppCloud is predominately a Gentoo-based custom portage tree that Ezra built years ago, and it has seen very few updates. This causes problems building gems or bugs in gems that are quite rare since most people do not use software/libraries that are so old.
While there may be amazing things in the works, it also seems to take a long time for features to be rolled out, and often these are in betas initially which do not fall under the support contracts. So if we wanted to move to Ruby 1.9, it's at our own risk (or was, last time I looked).
As for the jabs at Heroku, I have seen quite a few, honestly, but I don't have a pile of links for you. Many of them are subtle or implied, and not necessarily untrue. For instance, sure, Heroku was acquired by Salesforce, but I haven't seen any indication as to why that should influence my hosting decision. Companies are bought and sold all the time. The "ey-migrate" tool was also a great time to get a few subtle jabs in.
I am--was--an EY customer as well for some time, about a year, and had been following it since its inception, particularly through a friend who was a customer while it was still in beta.
1. I haven't seen any significant changes to their product, except for the CLI tool as a replacement for capistrano. The CLI tool is a great improvement, but it's still basically a glorified capistrano tool, and doesn't even get close to comparing to the Heroku command-line tool. Predominately, I feel like the product has been stagnant. Support for Ruby 1.9 and Rubinius (currently in beta) is the most exciting thing I've seen lately. It's surprising rbx has taken so long when it's an EY sponsored project!
Part of the challenge with a product like AppCloud is there is no obvious way to update the stack without disturbing applications, but I'd still have liked a way to potentially update our base software (libraries and such) to something more current.
2. I'm so-so on this. Heroku support is free, but is not 24/7 unless you are a very large customer with a support contract. We paid for support, and EY support was sometimes helpful, sometimes not.
3. I'm not talking about objective comparisons here, these are "shots across the bow" so to speak. I know the pros and cons of both services very well, but what I am talking about are "subtle jabs" about how Heroku is not production-ready/capable, etc.
What I meant by that was that we weren't giving away 500 hours on a micro or small instance. We've not found them very useful for Rails/Ruby web apps. It costs us more to give away 500 hours of mediums, but we think the experience is a lot better.
This isn't a jab at heroku. We had the option to provide micro, small, or medium instances. We host most of our apps on mediums so this is a good, powerful instance. 500 hours on a micro is nothing compared to 500 hours on a medium.
There are differences between each and every provider and every developer needs to choose which service they want to use in their own way. Customers choose Engine Yard for many reasons including 24x7x365 human technical support availability, a choice of infrastructure providers, resource transparency, request transparency (EY is not in the request path to your app), customizability and a strong history of supporting the Ruby open source community with projects such as Rails 3, JRuby, Rubinius and Fog.
We're very aware that there are other services available, and that they fill a need for certain users. Specialization is important, and it's difficult to be all things to all people.
Ah, I see! We knew that consulting shops would like this because it allows them to get an account set up without a credit card, and ask the client to complete the signup process, including handing over their credit card. :-)
Clients with limited budgets are easily sticker shocked by hosting costs. Having the option to bundle 20 days of free production hosting is a great addition to any development service. EY AppCloud is a great service and we hate to see clients opt out without giving it a try.
We considered giving more time on smaller instances, but in the end we felt our customers are more concerned with getting production sites up and running. So we're giving 500 hours of time on instances highly suited for public facing sites with sizable traffic.
No, we don't expect many will make a decision based upon 500 free hours, but we do expect many will give us a try, and make their decision based upon the experience they have during their 500 hour trial. :-)
Sure. I think it's a potentially effective promotion for exactly the reason you describe. (free test, and if they're not actively dissatisfied, switching costs will likely cause them to stick with you.) That's an idea worth testing.
I just can't understand the parent post indicating that it's 'huge for rails consulting companies'. After all, rails consulting companies are sending out five digit monthly invoices. A $150 discount on the first month of hosting is a rounding error in that context.
Ha, that's possibly true as a percentage of total fees. Hopefully the total fees paid to Engine Yard will always be a rounding error in the context of consulting fees! :)
We're investigating other benefits for our Development Partners too. Any thoughts?
So glad to hear that, let me (@tmornini) know if I can help. We also have a partner program for Rails consulting companies, you should check that out too!
AppCloud seems to be (and I hate to use this word) "just" a thin wrapper around EC2. At least I get that impression looking at the price points - the smallest 32bit instance is just shy of 1k/year. Am I going to get more performance out of three Dynos at Heroku for about the same cost? Maybe maybe not. But where am I going to put my staging test/servers? On a 20/month VPS? On an instance @AppCloud that I spin up and down? No, I'll tell you where, I'm going to stick it on a Dyno or two at Heroku and not worry about it. And then when I need to scale or the product is ready for production where is it going to be hosted. That's right - the same place it's been validated and tested on.
By the way, Engine Yard is not the only place that has this barrier to entry problem. Most of the cloud providers do. Joyent.com, Media Temple, Rackspace cloud all have Barriers whether it's poor documentation, higher than normal base costs and bandwidth fees or just inconvenience like having to wait on a call from someone in Texas in order to be allowed to create instances.
If the other players want to compete with Heroku on the service and ease of use side or Amazon on flexibility and scale then they are going to have to do much, much, much better.