alanwilliamson
The line between what is considered a cloud and what is paramount to hosting is a blurry line at the best of times. What qualifies a companies offering as a cloud based product offering? That is a question that is continually debated in the cloud circles and no one has really yet to offer a real concrete reason.
Setting aside all the marketing buzzwords that will get thrown in "scalability" / "on-demand" / "deploy", a cloud offering has to deliver on its hype.
For me when words like those are thrown around I want to see:
- Scalability
The ability to start up a new instance programmatically via an API, allowing me to hook into whatever trigger metrics I may have. So if traffic reaches a new level, throw more resources/machines at it without doing anything further except start it up. Also ability to turn off resources to scale-down. - On-Demand
When I do start-up a new instance, I want it to be fast. Do not make me wait an hour, or make me have to provision it before I can use it. If I can't introduce a new instance into the system in under 5minutes then its not "on-demand" - Deploy
Tools that make it real easy for me to deploy my software/services in real-time. I should never have to resort to an SSH shell to make my software work.
If an offering does not meet those criteria then for me, I do not believe they are truly offering a cloud experience. They are merely repackaging a hosting package. Running your machines on a virtual platform does not a cloud offering make. Just because a company is charging per-hour instead of per-month, don't be fooled. Look further.
Aptana is a company that is renowned for providing an IDE for building web applications, based on the Eclipse toolkit. They have started to pop-up on the cloud radar, that initially confused me, as I tried to fathom out just how they were making this claim. Though their Jaxer product allows you to run your Javascript server-side I wasn't sure how they were making the leap to clouds.
Looks like Aptana is moving into the hosted environment, which is a logical step for them considering their big play in the server-side market. What doesn't make sense though is claiming they are now suddenly a cloud offering.
They fail to deliver on my 3 step criteria, with the only scaling they seem to offer is within the machine itself (more RAM/HD as and when). No details on how long a machine takes to setup or run, or what happens when you need to grow outside of one machine. No load balancing or image provisioning details. No details on what happens when you need more disk space. No details on any API that can be used to create new machines as and when.
Don't get me wrong, but this is a great offering from Aptana and compliments their product suite perfectly. But it is far from a true cloud offering. If they had made more of a play like Google's App Engine, abstracting all the logistics away then yes, we could then put them into the world of cloud computing, but as it stands they seem to offer simply a hosting model for their studio.
Demand more from the companies that claim to offer cloud computing. Ask them precisely what makes them different from a traditional hosting company, and if they can't achieve the 3 points I highlighted, then just look at them as you would with a traditional one-server-one-client hosting company.
Remember, only one week to go until the free cloud boot camp in San Jose. We've got a lot of material to shoot through, so be sure to come along and get your head truly immersed into what some of the big cloud operators can do for you today.
Comments
please note, all comments will be moderated for spam and abuse before being publicly posted.
Article Details
- Published:
10:11 AM GMT, Tuesday, 11 November 2008 - Categories:
Technical General - Tags:
cloud computing aptana hosting - Comments:
4 left; add comment
Related Articles
Article Archives



Christopher, thanks for you comment.
I think what you have is a wonderful deployment story, but as for scalability, I do not believe you do scale. Giving a server more memory/CPU isn't scaling. What you have is a high end "hosting" offering. What happens when you do really build that ultimate Facebook app that needs a huge infrastructure to support? Aptana will not provide that. Even Joyent, your underlying provider, requires some hand holding to manage that process. After speaking indepth with Kevin Hakman from Aptana regarding this, I feel Aptana's play here should be more "Google App Engine" and less "Amazon EC2/Joyent". You have a beautiful end-2-end story from code building to deployment, but to say its a cloud, I believe is not quite correct. What you have is a wonderful hosting story. Or put another way. What you have today, you could have provided years ago. You weren't waiting on cloud technology to make it happen. Aptana is a first class tools development platform, without a doubt. But a cloud provider? No I don't think that is your story. Hope we can catch up today or tomorrow at Cloud ExpoI think that Aptana's Cloud does actually meet your three criteria, though not as fully as you might hope. The cloud sites are scalable, but right now they are only vertically scalable - the site can be scaled up or down on the same box (or migrated to another box), getting allocated more or less Disk/RAM/CPU. Horizontal scalability, redundancy, and load balancing are the next major step we're looking towards. I don't know that an API is really a concept tied to Scalbility, but what the IDE does is just call an API to spin the box up. We haven't yet discussed making the API public, so right now you spin sites up inside the IDE, though there's no technical reason we couldn't make the API public, and we do have plans to eventually get a web site that provides much of the same functionality as resides in the IDE. It is on-demand, with the sites coming online in less than a minute, typically. From my last timings our average is around the 40 second range. I think the scaling timings also matter, but that's a longer discussion. Lastly deployment - we offer an integrated solution with our IDE to "deploy" - it uses an SFTP client to compare local and remote and performs the diff. We also support other SFTP clients, using SSH and SCM as a deployment, or using something like capistrano to deploy.
Actually I think Aptana Cloud currently meets two of your three criteria. Starting up a new instance or changing its capability is very fast. Also you don't have to use the shell at all in order to get a typical web application running.
We are working on an API. We are also working on distributed solutions.I think this naming problem is going to roll and roll, very much like the web2.0 tag.
I'm not really sure it matters too much if a provider claims to offer cloud like services but doesn't really. If these packages were aimed at the consumer market, we may end up with some confused individuals being miss-sold products but ultimately as developers we should be aware enough to ensure we select the correct product for our needs. I see the whole hosting industry moving to try and adopt the three criteria you state above. It simply makes good business sense. Of course, whether that API call triggers an intern to go switch the box on or not will probably differentiate the key players from the hangers on. But then the hosting industry has always had hangers on, as to whether these can continue with the aggressive pricing models begin offered by the larger players (thinking Amazon).. probably not. I saw you speak at the AWS Startup Event in London last week. Very much enjoyed your presentation. Thank you.