In Cloud Divided Part One we examined the developing divide within the elastic hosting space between large public wholesale hosting providers such as AWS, Rackspace, and Google App Engine and smaller hosting firms run by teams of primarily veteran engineering talent. In Part Two we will take a look at 5 examples of companies that are leveraging the large public wholesale hosting providers in very unique and powerful ways; Iron.io, Soasta, Engine Yard, Heroku and DataPipe.
Iron.io’s platform leverages a seamless integration with elastic compute environments such as AWS, Rackspace, Joyent, etc. to let you outsource specifically the managed execution of your code. Nutshell, the threads (workers) you can have working on a process are not limited by the hardware, and thus jobs which may take your existing environment minutes, hours or days can now take seconds or minutes. THIS, in our opinion, is leveraging the real value large public wholesale hosting providers actually provide.
SOASTA provides a scalable web and mobile application QA and testing platform leveraging integration with elastic compute environments around the world to simulate thousands (or tens of millions) of users hammering an application from anywhere in the world. Over the past few years they have become one of the leaders in the industry offering something no other software/emulation based QA and testing solution can today. Again, THIS in our opinion is leveraging the real value large public wholesale hosting providers actually provide.
If you are a frontend developer chances are you have heard of Engine Yard and Heroku. For those not familiar, Engine Yard specializes in managing ruby, rails, and php, while Heroku also supports java and python. These companies provide a managed hosting solution running on top of primarily AWS’s infrastructure, meaning they are reselling you the same compute power you could receive going direct to AWS… though packaged and marketed under different names. You pay a premium however working with Engine Yard and Heroku for the ability to access their engineers and custom built solutions which ideally help you scale more effectively.
Datapipe is a private colocation and hosting provider with facilities in the USA, London, Hong Kong, Shanghai and Iceland (yep… Iceland… kind of crazy, but we’ll discuss some other time). What’s interesting with these folks is they not only provide the traditional datacenter and private or shared hosting solutions customers are familiar with, but they can also help you manage the applications that you feel are better suited to leverage an elastic compute environment. This type of environment, known as “hybrid cloud”, is pretty awesome IF it can be managed effectively. Reality, however, is that as sexy as this solution sounds, it would never be needed or recommended (at least by us) for 95% or more of businesses out there today. In that it’s nice to know your provider CAN help you with such a deployment should you want or need it down the road, our hat is off to DATAPIPE for being one of the more progressive traditional datacenter + hosting providers we’ve come across.
In conclusion, companies like Iron.io and SOASTA represent a general evolutionary trend in the elastic hosting industry that we are excited to see more and more of as it leverages technology to provide solutions in ways that simply didn’t exist just a few years ago. If your business is determined to leverage an elastic hosting provider however, unless your management has mandated you work with AWS, Google or Rackspace and you have the in house expertise to scale within such an environment, our experience has taught us that you will receive FAR better service, more visibility into the backend infrastructure of your environment, more control over tweaking this environment, and in most cases better pricing when you work directly with the team responsible for managing your application AND the infrastructure itself. We also feel DATAPIPE offers a forward thinking solution that could be ideal for the client looking for help managing across multiple deployment types.