My suspicion is that virtualization is really just a transitionary technology that gets us over the open & concurrency hump rearing up in the technology road ahead of us.

It’s not that this hump poses a problem to everyone, many developers are well ahead and have been riding it’s crest for some time, building concurrent and or multi-tenant solutions that run natively on the internet.

But organizations of all sorts have a great deal invested in pre-hump technologies, a transition to post-hump is going to take time. In the interim, cloud vendors, shower us with their virtualization rainbows to lead us to the open native internet pot of gold on the other side.

So what is at the end of the rainbow? Although its difficult to see (primarily due to the cloud marketing fog obscuring it) one would imagine it looks a lot like the internet does now only bigger and better. That is it will be built around open standards like TCP/IP, http, html, xml, javascript, ajax and perhaps some newer emergent ones. We should also expect to see REST rather than WS* hairball, view source rather than binaries and most likely OpenSource rather than proprietary. Whatever is there, it will be used to build the new native internet software and services world post-hump.

What is fascinating is what happens to the current diversity of native web production which emerged from CGI grew up through LAMP evolving into the popular state of the art frameworks like RAILs, DJango and cornucopia of similar solutions in other languages. We already see emergent post-hump candidates like Google’s App Engine and Microsoft’s Azure tools + platform, can these succeed or will diversity rein, perhaps running on more diverse plays like Amazon’s AWS or a Eucalyptus fork. I personally have a difficult time imagining any single one of these dominating, I would expect diversity to continue but with some consolidation. I also have an inkling that something else is likely to emerge soon that changes the game around concurrency, something that makes it central to its architecture. In many ways it will be an anti virtualization force, it will also likely be a highly efficient development environment and solution consuming less and delivering much more bucking the historical trend to date, it really would be development gold!

What do you think might emerge, have you seen the signs yet, anyone got some clues to throw us?

I’m not alone in thinking that Google may do things differently with their cloud offerings, Winer thinks they will go with the freemium model (free and payed premium options) unlike Amazon’s AWS which is pay only. Freemium makes a great deal of sense for Google and is a model they obviously like, just take a look at Google Apps, it follows the freemium model perfectly. Google also has a history of doing free first and finding a business model after. In terms of business models there are a number of suggestions. Winer suggests reduced cost for Google technology acquisitions, the cloud acts as a combined VC/Technology platform, thus when they acquire successful implementations on their platform, integration into Google becomes a breeze. They also gain mindshare allowing educational establishments to use their cloud for free, becoming part of the curriculum. There is also a huge identity battle going on right now with Microsoft,Yahoo etc.. this move could deliver a fatal blow to the other identity players if the Google cloud use became prolific. In addition successful implementations would likely move to the premium GCloud, bringing in direct revenue. Ok enough speculation around Google’s motive and business models lets crack into how the hell they could do it.

Now if we assume they go the freemium route, they would need a system that can scale with free. This is no mean feat, free can get pretty prolific very fast, if free meant everyone got an EC2 equivalent that could add up to a severe drain on even Google’s resources. This is where I believe Google may ‘think different’, they will need more control and greater scalability without limiting innovation. Also I believe they will want to make it easier to get started than the current AWS offerings, they would need a shorter learning curve. In this sense the GCloud would need to have some basic constraints and familiar development models to attract modern web application builders. So how will they do it, what will the cloud be made of, how will regular web developers use it?

So here is the really speculative part, if we assemble together Steve Yegge’s Next Big Language (NBL) and his recent project G-ROR (No not Ruby On Rails, it’s Rhino On Rails) as well as his top secret project NBE (Next Big Environment?) we can pick up some important clues. This are useful for several reasons:

1. Google intend using this stuff internally to build apps (at least prototyping them)
2. It is a perfect vehicle to build google API infrastructure into (GData,Search,OS..)

So what is NBE? My guess is that NBE is a combination of Java Virtual Machines (JVM) plus supporting Google API libraries with security wrappers. Here is how it could work; Google’s equivalent of EC2 would be a Cloud Virtual Machine (CVM), this CVM can run any suitable Java byte code that has been built with the CVM toolkit. The CVM toolkit does some sanity checking etc..to prevent abuse. Each CVM acts like a secure independent system (note – not operating system) isolated from the others with controls to enable you (or Google) to start, stop and control it remotely via a CVM API. These CVM instances will of course run in the Google managed datacentres.

So how will developers build the CVM apps? To start with I believe they will use G-ROR, a rails like development platform that includes an equivalent to active record for GData, allowing data to be stored in the GCloud transparently. It will be a development model that many are already familiar with, one which has already shown itself to be popular and very quick to get up to speed with. The main difference will be that the programming language (initially) will be Javascript rather than Ruby, this will appeal to an even wider developer audience, Google loves javascript in case you haven’t noticed – NBL?.

So there you have it, that is how they could deliver cloud computing ABC economically and with the constraints/controls required for a freemium model. They could run numerous CVM instances per server to make the whole infrastructure scalable and financially viable.

I think they could also offer premium CVMs, with extra features (more memory, dedicated processors etc..) building ontop of the JVM/CVM. They could for example offer existing server side language choices from the JVM such as JRuby, Jython, Groovy and Scala to name just a few. All of these compile down to Java byte code and could thus run on the CVM platform just by adding the required support libraries. And guess what, that would make Google’s CVM one of the most language accessible cloud platforms around, just look at how appealing that could be to the web developers everywhere (NBE?).

Of course this is all just idle speculation, but there may be a grain or two of truth to it. Does this Google cloud offer a developer’s silver lining ? I would be interested in what you think, does this idea appeal to you as a developer? Is what I am talking about severely floored? I am completely
nuts?,I wanna here you thoughts, go on let loose…

*Note this post was originally posted here