The road to cloud computing continues. Last month, we asked what are the first steps in moving to the cloud.
This month, we want to get more specific and hear your stories. Remember, the best comment wins a MacBook Air. Our question:
How do you move applications to the cloud?
Sponsor
Leave a comment on this post with your thoughts and we’ll review it for judging.
My co-judge this month is Sam Ramji, vice president of strategy at Apigee. Sam is a leader in the use of APIs. He has a clear vision for how applications integrate and exist in a cloud environment.
Now on to the winner!
First off, a big thanks to my co-judge George Reese. George is the founder of enStratus, a cloud infrastructure technology solutions company with customers that include Korea Telecom.
George and I want to congratulate Joe Masters Emison, a vice president at BuildFax
http://www.buildfax.com/, who clearly explained why his company moved to the cloud and how they did it.
Here’s what he had to say:
I run research and development for a data company, and have overseen the transition of our production data-processing environment from local servers to virtual servers on the cloud. We decided to move to a virtualized infrastructure as a result of our experience with application failures on congested servers.
When we decided to move, the most important considerations were (1) how long is it going to take us to verify that our planned cloud deployment will work, (2) what connections are we going to need to make between our existing architecture that we are not moving right now, and (3) how long is it going to take us to deploy fully on the cloud? Understanding the timeline of deployment was critical for us to determine what stop-gap measures (both in terms of hardware and personnel) and what infrastructure changes (in terms of communicating with the cloud-based servers) would be necessary to buy us the time to move.
The first steps in deploying, for us, were to take the smallest kernel of what we needed to move and test it on the cloud. We have a data-processing pipeline for which each step requires varying amounts of human and machine power, and one fully automated step, the “importer”, was the heaviest user of our local servers. So we launched a virtual server, and tested the importer application on that virtual server with sample input. That worked, so we built out connections to the virtual infrastructure: (1) so that we could launch virtual servers on demand, and (2) so that we could send input to the virtual servers and receive output back from them.
I have found that the most effective way to deploy a simplified, reliable, and optimized virtualization solution is to do one thing well, and then expand from there.
That’s a straight forward explanation that goes into a bit of detail about the deployment. We’re still waiting to hear from Joe to confirm his prize. If we do not hear back from Joe by Friday then we will pick a new winner.
Criteria
Our criteria for choosing the winners is as follows:
- Depth of understanding
- Factual examples to back up assertions
- How the conversation develops
- Accuracy
- Strength of argument
- Concise writing
- Originality
The contest will run for the next 30 days. The winners will be announced on the first of April. A Disqus account is required to participate. Please take a few moments to read the full list of rules before joining in the conversation.
We look forward to the conversation!
0 Responses
Stay in touch with the conversation, subscribe to the RSS feed for comments on this post.