For w while now the buzz is about being cloud. Whether it is apps in the
cloud, services in the cloud, or being a cloud provider you can’t seem
top escape this buzzword lately. But given the recent releases and
direction the “major players” are talking about, I have to wonder if we
are entering a stagnation period. Or perhaps, to borrow a phrase from
economics, perhaps stagflation.
First, this is what I see happening.
Start with the vendors. In this group I include Amazon, Rackspace, Google, Microsoft, et al.
What are they doing? They are trying to be “cloud” by making ordinary services available “in the cloud” and call them cloud services. From “Database as a Service” to “Platform as a Service” or “NOSQL Database as a Service” they are trying to turn every product into a cloud service. But I question whether running MySQL in the cloud for you constitutes a cloud service. There is no (as of yet) MySQL cloud database there is just MySQL run on cloud infrastructure. There is a difference.
Let us take something I know a fair bit about: Redis. Every provider of “Cloud-ified Redis” today is simply providing a Redis service running on a platform they manage (or someone else manages for them), or managing Redis for you. Neither of those are “cloud”. Those are standard, decades old managed services model with a new buzzword thrown onto it. So, what is different from managed services to qualify as innovation in the cloud space?
Imagine you don’t care whether it is Redis or Memcache. You have a single endpoint and single protocol. You don’t provision an instance. When all you need is an API and/or SDK to integrate it, that is when you are a cloud service instead of a service on a cloud. To apply this to databases imagine having a “Cloud SQL” library where you configured it to talk to an endpoint and went from there.
So perhaps some of these don’t quite work that way. That does’t mean simply removing some management from the person paying for it qualifies it as innovative - we were doing that about 15 years ago.
Further, many of the “as a Service” offerings are not built to be deployed in cloud infrastructure. They were designed for the era of local disk, or NFS, and single head interaction points. For an example, look at the PHP CMS and Blog engines available today. Put them in front of a dozen web nodes running NFS to keep data in sync and available (not to mention make the web front ends replaceable) and they have all sorts of issues as they try to cache objects, code, data, and the HTML returned.
In this case the implementation language has issues with this concept (it wasn’t designed for concurrency), but also suffers from assumptions made by the implementers of the platform - such as the idea that to scale horizontally is easy and something you do at the OS level. Nonetheless, this doesn’t change the fact of these platforms not scaling well - if at all.
Thus, my tentative assessment that we’ve stagnated in innovation since all we are doing is adding to the product catalog by taking existing products and running them in a semi-automated fashion “as a Service”. We as a community are inflating our product lines, tacking on “as a Service” and calling it cloud innovation. Hence, stagflation.
I suppose I should call it “Cloud Stagflation”.