This is close to what we'd want to replace the business model of a bunch of sketchy startups.
Startups with nice "todo"/personal data apps, etc. (cases where it's silly for the user to put their data in strangers' hands for no apparent reason) could "sell" them to users. If the startup disapeared, the user could still pay for web-app hosting.
The problem with using this particular AWS offering for that purpose is that every app would cost you hourly to run and would vastly underuse the computing resources that the user is paying for. It would be great if there were a similar offering based on a programming model where many apps could be provided by a single instance, or even perform on-demand provisioning.
I think this is the future! Somebody make it happen.
pageforest.com? (my project). But ... no payments system built in. I found that the JavaScript clientside-only programming model was hard to get developers to adopt.
Looks promising, particularly for the Todo-list type webapps which I suspect could be implemented to run entirely on the client. Are there any apps in particular that you'd recommend I look at to see what your platform can do?
I think a full solution will allow people to implement server-side functionality as well. Perhaps that's in your plans? Also, I think the ability to sell access to apps would help attract developers to your platform.
That's cool. However, people must still understand the software they are using. For instance, someone is selling Nginx 0.8. The latest version is 1.0.15.
Agreed. I saw "turnkey rails" in the sidebar and thought I'd take a look into it. It's very unclear what versions of software are included. The last changelog entry on 2011/12/06 says nothing about ruby or rails versions. The last mention was 2011/01/27 claiming REE 1.8.7-2010.02_i386_ubuntu10.04.
To give a perspective on update frequency, according to the same page (http://www.turnkeylinux.org/updates/rails), two updates before that was Fri, 2009/10/23 - 16:35. This update claims to have ruby 4.1 installed. It seems that their VM image also includes a time machine.
Not getting a huge feel of confidence from that being featured on their landing page. An open market can be good, but when it comes to supplying an image for my production env, that type of thing could leave me feeling a bit uneasy.
It's a shame, because this is how we get massive security scares that affect the whole industry. One person couldn't be bothered to upgrade Apache and the whole industry is insecure.
Nginx 0.8 with support. Sure, 1.0.15 would be better, but enterprises generally care more about support than about having the latest version (see Red Hat).
Version numbers aren't that relevant for Red Hat's packages, because we backport security fixes (and features) into the older versions. And yes, customers prefer it that way.
Through the vulnerability scanning I do at work, our tool marks a lot of things just based on version numbers. Looking into them deeper, I see the backported patch has been applied. RHEL does a pretty good job on this, really stands out. Makes my vulnerability scanning a little more time consuming, but frees up time I would otherwise be spending on documenting security breeches and data loss.
Ok, I don't really have a problem with how Red Hat do things. However, if I was employing a sysadmin to administrate my servers, I want the latest production software please.
If you're worried about security updates, Red Hat monitors dozens of channels and backports all security fixes to all supported software, so that's not a problem.
If you're worried about features, Red Hat selectively backports features at customer request, or gives the customers free updates to the new RHEL.
Our customers want to install a particular release of RHEL and have it work and supported for 10 years. They actively don't want random version upgrades (which often remove as many things as they add, as well as causing exciting new bugs).
I wonder if companies like mine could proxy/resell for you, since we have a US subsidiary in Boston that meet the criteria. We're working on rolling out our own issue management product into the AWS Marketplace and having a local US office was a lifesaver for us, for this and other distribution and billing bureaucracies (related to consulting in this case).
We're not talking about a bootstrapped startup though; this is Amazon with all the legal expertise that means. They managed to evade £9 million in UK tax!
Just because Amazon is big doesn't mean that everything they do has to be big and full-featured right from the beginning. That's how big companies ossify.
I guess one could always find a justification. But the fact of the matter is that no other big US company managed to do what Apple did years ago: give money to non-US companies/individuals selling IP.
Google's AppStore still supports only a limited number of countries.
Amazon don't support sellers outside the US for any of their offerings (be it, selling books via Kindle Direct Publishing or selling apps in their Android store or selling EC2 images).
This is not a feature that's too early to talk about because AWS Marketplace is new. Supporting non-US sellers is something that Amazon just doesn't seem to be able to do it or doesn't care.
Frankly I don't know if it's a matter of red-tape (they would need a single EU presence, for example, to cater to the 27 member states) or just some kind of weird protectionism.
I worked at Amazon for several years; one of my team's biggest projects during all those years was expanding certain partner programs to new international locales.
I think Amazon generally prefers to launch quickly in one or two big locales, then expand geographically over time. The alternative is to delay the launch everywhere until support for every region is ready. Launching everywhere at once has its benefits (in particular it can make for better publicity), but potentially at the cost of being later to market and the opportunity cost of revenue not earned during the delay. Amazon has less focus on major PR events than Apple, so it probably weighs these costs and benefits differently.
I wish Google had a better version for AppEngine. One click provisioning and integration with Apps domains for authentication and user info with a SaaS style payment processing built in.
One advantage of the image-based model of the AWS marketplace is that you can support nearly any type of product and multiple operating systems, so you can adapt existing products rather than have to write your app specifically for AppEngine
A bit different. What I would like to do is instead of provisioning users on my own appengine instances, give them a button where they can sign in with their own Google accounts and deploy the same code to their own instances, where they can then take care of their own billing, integrate with their own users, etc.
Add to that a way to customize the price so that in addition to the appengine billing the developer gets an extra $x per month for app licensing
My point is more general than that specific stack, the concept of one click deployment stacks which are built by Sys-Admins and have support is a big deal.
I wonder if they will allow multiple apps sold from the marketplace to run on the same instance? I haven't gone through the process yet of setting up an instance via the marketplace, but I think that it's essential that I can a.) install multiple copies of the same software on an instance and run them via virtual hosting (for example running multiple Wordpress instances on a single node and b.) allow different kinds of software to run on a single instance (let's say an instance with a bug tracker and a log analysis tool, just as a made-up example).
The other thing that I find interesting is that there doesn't seem to be an API yet. I would love to be able to provision apps on nodes via an API to the marketplace.
I'm definitely going to play around with it a bit and see what's possible though.
You can do this with the Bitnami instances available in the marketplace. You can launch one of the LAMP-based ones and then download modules from bitnami.org and install them from the command line
It's good to see this coming. However I don't understand the price structure some vendors are offering. Take Zend Server(below is the link) as an example, why should they charge a lot more on high-memory/high-cpu instance? Is it because the incurred traffic or usage is much higher? How does this compare to setup your own server but buy server license from them.
The typical enterprise pricing approach is to try to price according to their value, not to your costs. If you are willing to pay up for a bigger server, you are probably getting more value, and therefore should pay more.
The only thing that struck out at me as odd from this announcement was that it wasn't phrased like: "Amazon is excited to announce!" (Most of you probably know what I'm talking about)...
This is the firsteverever product announcement that I've ever seen Amazon make where they weren't excited to announce.
> Amazon Web Services is pleased to announce AWS Marketplace, an online store where customers can find, buy, and quickly deploy software that runs on AWS.
We’re excited to see that you are excited about our excitement over our exciting series of releases.
If we change to another adjective, the only logical post title would be “Amazon is no longer excited.” And that wouldn’t be too exciting now, would it?
It may be a Seattle-area thing. When I worked at Microsoft, the "super excited" phrase was a bit of a running joke, but it still didn't stop us from using it at every conference, in every public e-mail, hallways, etc.
Search for Microsoft AND "super excited" and you will not be disappointed. You will see both modern hits and results dating back to the dark, dim days of web indexing, from which results are rarely returned.
This time s?he was simply pleased. Maybe s?he doesn't like this one.
Amazon Web Services is pleased to announce AWS Marketplace, an online store where customers can find, buy, and quickly deploy software that runs on AWS.
A lot of us get those emails from amazon.com, like the one this morning, "Introducing AWS Marketplace -- Find and Buy Software that Runs on the AWS Cloud". This is great, but I have a friend who likes reading about these but doesn't want to launch servers. Is there some way people can subscribe to the Amazon AWS mailing list without creating an AWS account?
I guess RSS is the best option. Even if you don't have to launch servers I still believe you need to confirm your phone number and some other things. I'm just looking for a way to share news with friends without having to forward the e-mails every time.
It hasn't held them back. I wonder if it's a case of their services been so unique, cheap, competitive, etc. or if we overrate design and it's not really that important. I tend to think it's the former, but it's an interesting example.
My issue isn't so much with design but rather the fact that you have to register a new account with them every time you want to use a new service. This is almost certainly legacy complexity from the days that these services were not exposed through API's, but it is still a pain in the ass.
I'm confused. What's the difference, other than Amazon taking a 20% cut of my sale price, between this and the already existing paid AMIs? Yeah, there's a pretty website and all, but it doesn't seem like much value for 20%.
The listing of Linux images including EC2 charges is scary. I hope this isn't their long term plans. If so, we'll have to start coding our way out of their vendor lock-in.
Ruby could be obfuscated by compiling to a jar using JRuby. Presumably something similar could be done with Python and Jython, and PHP can be compiled into C++ using Facebook's Hiphop compiler they open-sourced.
Why it is irrelevant? I want to sell my application via AWS so users can deploy my AWS image with my application by I want to protect my IP. How to do it?
Haha, I appreciate the sentiment. And yeah, Amazon's marketplace is pretty exciting for us as well - their approach is obviously quite different, but it's very cool to see movement like this higher up on the stack.
AMZN is up about a percent, which is not at all unusual and can hardly be attributed to them launching Marketplace. Hell, they were up that much yesterday.
Startups with nice "todo"/personal data apps, etc. (cases where it's silly for the user to put their data in strangers' hands for no apparent reason) could "sell" them to users. If the startup disapeared, the user could still pay for web-app hosting.
The problem with using this particular AWS offering for that purpose is that every app would cost you hourly to run and would vastly underuse the computing resources that the user is paying for. It would be great if there were a similar offering based on a programming model where many apps could be provided by a single instance, or even perform on-demand provisioning.
I think this is the future! Somebody make it happen.