Since the dawn of Cloud Computing, there’s been a lot of talk about interoperability. Much of this has included discussions around working groups on interoperability and plans for open cloud standards. While this is well-intentioned, and portability across clouds is a noble goal, but practical demonstrations of cloud interoperability have been few and far between.
Many vendors have demonstrated their individual cloud’s capabilities, but for the 2009 Enterprise Cloud Summit we wanted to do something different.
For the last few months, Syntenic and Bitcurrent have been working on a sample cloud application designed to showcase the power (and limitations) of cloud computing. The resulting application, based on the Panda open source project, lets users upload and label video content that is then transcoded into a variety of formats. Put simply, it’s like Twitter with video.
We chose this kind of application for several reasons:
- It’s written in Python, a format supported by many cloud platforms
- It’s bandwidth-intensive (due to potentially large video uploads), processing-intensive (due to the amount of video transcoding required), and bursty (potentially streaming multiple videos at the same time), making it an ideal candidate for on-demand platforms.
- Many people can access the application concurrently.
Six cloud computing companies have devoted their time and resources to building demonstrations around this core application, from the underlying platform to provisioning to testing and development. We’ve also invited one company to show how its services bring legacy enterprise applications — rather than our reference application — into cloud environments.
- On Amazon EC2 we spin up multiple instances of machine images, each of which is a transcoding server.
- Using Rightscale, we launch front end instances that are managed and load-balanced.
- With SOASTA we test the application under load by generating synthetic testing traffic..
- Deploying 10Gen’s MongoDB, we show how cloud-like infrastructure can be used on-premises when enterprises have data that cannot be stored in the cloud.
- Using Google App Engine, we show how clouds interact – with a front end running in Google connecting to a datastore in the Amazon cloud.
- Finally, using Elastra we show how enterprise-standard applications including licence management can be built in the cloud.
- Update: Unfortunately, Elastra are now unable to participate. For our final session, we’ll be showing with SkyTap how you can perform traditional development in the cloud by creating a library of cloud machine images
Building the application on these different platforms, and testing it, has shown us a lot of wrinkles in the promise of clouds, and a lot of opportunities for improvement. It’s been great to get our hands dirty with some real-world integration efforts and dispense with theory for a change.
We’re going to demonstrate all six integrations as a series of short live demonstrations at ECS09, which is part of Interop ’09. (That week is a whirlwind of cloud events, including CloudCamp Vegas, the Cloud track, and Unconference.)
The feedback we’ve had on the application has been excellent. As a result, we’re talking to several other players in the cloud computing world about integrating our cloud application into their platforms and services.
You can download an overview here (PDF, 240Kb) and if you check back here after the demos you will find full details of each of the demonstrations.