A continuation of a short discussion between @jlivens and @Kloud_Karl about on-demand cloud services.
Giovanni Tropeano: So, just giving insights into what an architecture might look like, if you were leveraging the sort of services, talk a little bit about what that would look like?
Jay Livens: Well, what’s interesting is that, as customers think about moving to the cloud and leveraging the dynamic nature and flexibility, they’re really thinking about, well how can I develop my applications differently? How can I more rapidly create the applications that deliver business value, that can leverage the scale and the flexibility that Azure brings and increase the cycle times. Because the reality is that in past years, we might spend weeks or even months on every release. Today, it needs to be much faster. We need to be able to turn our applications really fast and monetize them in a more efficient way, so we can get more money and better services faster that if a bug or new feature is desired, we don’t wait nine months for that. We develop that, we implement that and then we move on to the next one. It’s really common as part of this whole DevOps initiative.
And a key element of that of course is leveraging the cloud for flexibility. But, it’s also having the data earlier in the development pipeline, so that you can develop your application on data that’s reflective in production versus some kind of subset or some idea that doesn’t really reflect reality, because if you don’t do that, you’re developing a subset, by the time you get all the way to the end of the cycle you find, “Oh-oh, maybe I didn’t develop it in a way that scales effectively.” And so now, all of a sudden, I find out at the very end of my testing, I made some very critical errors early on or assumptions that weren’t right and now I have to go back to the beginning, which significantly lengthens my development time and gets away from the benefits or the flexibility you can give.
So it really thinking about how you can expose data earlier, how you can get better code quality earlier in the pipeline, so that when you get to the very end, you really are just finding tweets here and there versus finding major problems. Karl, I’m sure you’ve seen this all the time yourself in your experience. What I’m sure you have some colleague or two.
Karl Rautenstrauch: Yeah, absolutely. Time to deploy, time to build has really changed in the age of continuous deployment, continuous integration and one thing that helps is — companies are iterating quicker on their applications and on their own IP, is having the ability to quickly provision multiple copies of that production dataset. To your point Jay, where you have to test on production data now, you have to test on that scale, you have to test with a valid record set and there isn’t time to go through the traditional modes of, well, here is my early Dev work then I’m going to go into QA and scale testing. It’s all part of the same process. So, multiple copies of the actual dataset provisioned quickly, available to the developers who are building those different and disparate streams. It’s really the only way they can take advantage of those “on-demand” and “rapid to deploy” environments that are available on the cloud. They need that same ability with those datasets.
Jay Livens: Agreed.
Access part 1, the previous On Demand Cloud Services article.
Next Topic – Leveraging Cloud for IOT
Webinar: Protecting Workloads in Microsoft Azure