PaaS is lacking a vision on application development
Platform-as-a-Service (PaaS) is all about abstraction and automation. Abstracting away from underlying technology layers by automation. That’s basically what is happening on each layer of a cloud architecture, from hardware to IaaS, to foundational PaaS, to aPaaS.
All this abstraction and automation is aimed at making application deployment a one-click or one-command experience for the developer. It makes deployment a self-service experience for the developer, which eliminates hand-offs and thus improves the productivity and scalability of the engineering team.
The disillusion that will follow the current PaaS momentum
PaaS technology is having a lot of momentum in the current market. For good reasons if you ask me. However, I’m afraid that this momentum will be followed by a disillusion…
Expectations and promises are high, so high that they never will be met by focusing on just a part of the application lifecycle. Just improving application deployment will not lead to the agility and productivity improvements that we need in order to support todays business demands. We also need to take application development and evolution into account. That’s exactly what’s missing in the average PaaS approach: a vision on application development. Programming is still as sluggish as ever before.
Abstraction and automation for application development as well
Why don’t we apply the same principles, that we so successfully apply to deployment, to development as well? Why don’t we explore abstraction and automation for application development more? Apply abstraction by using higher-level languages to develop applications. Apply automation by interpreting these higher-level languages or by generating code from it.
There are numerous reasons to do so, but the short summary is that it improves productivity, agility, and it allows to focus on the business problem. What is holding you back?
4 Comments Added
Join DiscussionHello Johan,
Since this morning I’ve been reading up on your blogs on MDD and PaaS and I must admit that you make a very impressive case. However, in all cases one strong premise holds true: developers (of any type) must either conform all their usecases to the harness of the model or not use MDD at all. I ran into these restrictions a couple of times and I’m strongly reminded of this article by Jimmy Lin:
http://arxiv.org/pdf/1209.2191.pdf
Loosely interpreted:
Map/reduce is a hammer, lot’s of algorithms are screws, so just reinvent them as nails!
My mind is still not made up on the matter. How does your vision compare?
Hi Rene,
In my opinion you shouldn’t try to do everything with the same tools or platform. The whole idea of Model-Driven Development (MDD) is to use domain-specific notions and generate or interpret software from them. So, all tools in the MDD world are focused on a certain domain. A certain problem fits in the domain or not, which should drive the tool selection. Or in other words: if a specific MDD tool is your hammer, carefully select nails, don’t hit screws.
[…] there. We should go beyond just optimizing how we deploy and operate applications. In my opinion the PaaS world is lacking a vision on application development. There are some hopeful signs, […]
[…] being perceived as more mature is that it is delivering business value quicker. I wrote before that PaaS is lacking a vision on application development. The driver for PaaS adoption is the need to be able to innovate. Companies need digital innovation […]