Anyhow, we do have external environment factors (the cloud providers) that will seek to mutate the DNA of the cloud for monetary reasons. Thus, we will have many many yet another framework for the cloud. Ultimately, I think the evolution process will make things survive or die. Those stuck with a dead framework could die with it.

Looking back, I agree with your history, with one interesting discontinuity between the progression of app servers and cloud servers.

Stage 2 app servers included NetDynamics, Kiva (both of these became iPlanet in some form), Netscape LiveWire (server-side Javascript baby!), ATG Dynamo, BEA’s M3 (the CORBA-based Tuxedo), and Microsoft Transaction Server with IIS. It was like the wild west.

The unthinkable happened in between stage 2 and 3. Let’s call it stage 2.5. An application server standard emerged with the Java platform. The combination of JSP, Servlets, and JDBC were enough for most people, but the options of EJB and JTA for transactions in early 1998, and the packaging format that J2EE brought in late 1999, meant it was basically the main game in town. For almost four solid years, 1998 through 2002, it grew at an unprecedented rate in the enterprise, and basically trampled MTS/COM+ and other proprietary app servers, until Microsoft finally fought back with .NET’s announcement in Summer 2001 & release in early 2002. I watched Wall Street shift from VB6, C++ and some dabbling with MTS/COM+ into full fledged Java engines during this time… .NET (and its push to Web Services) came in the nick of time to change the game. In the meantime, BEA had become a billion dollar company, and IBM WebSphere had a massive contribution to IBM’s middleware sales.

The moral of the story, however, is this: if my memory serves, the reason the enterprise Java APIs and J2EE were so successful was for one reason: the customers *demanded* an app server standard, and knew exactly what features it wanted: database access, web page creation, transactions, messaging, and a packaging standard. The bits of J2EE that failed were the stuff no one asked for (e.g. EJB Entity Beans in 1.x).

The problem with cloud seems to be that the enterprise customers have been very lax on demanding standards so far, or even understanding the benefit. The web’s benefit was clear: more distributed, easier to maintain, thin, non-Microsoft-dependent information sharing and UIs. Java’s benefit was clear: more productive server-side development. Cloud’s benefit is a confused mix of “The New Outsourcing”, “The New Agile Operations”, “The New Virtualization”, and “The New Massively Parallel Processing (MPP)”. Those are very different problem spaces, with (arguably) separate standards needed.

Great post. As a veteran of the app server wars, I find the parallels interesting.

First, I hope I understand what you mean by ‘Cloud Framework’. If so I think you would include current players such as 3tera/CA, Elastra, Rightscale, Eucalyptus, and not Cloud Operators such as Amazon and Rackspace?

My comment is about the benefits of “winning” the Cloud Framework war, in that I don’t see the same amount of upside as there was for the app server war. I worked for BEA, which ended up with revenues north of $1.5B selling license+services for app server and derivative products. Over $500M of that was license, and a big chunk of services was mandatory maintenance tied to that license. Its tempting to look for the same revenues to come from Cloud Framework.

But its getting awfully hard to sell licenses for framework software these days – e.g. no one pays license to run the likes of Spring, GWT, RoR, Django, Wicket, Seam, etc. Of course there is always money from optional support, but when it isn’t mandatory it doesn’t flow nearly as much.

While most of the current crop of Cloud Framework vendors have a license model, its hard for me to see that continue. Perhaps the open source devops stuff (Puppet, Chef) will take over Cloud Framework, or it will be Eucalyptus, or perhaps something else entirely.

The point is, winning Cloud Framework will certainly bring in millions, but I doubt it will bring in the billions* like winning app server (viva WebLogic). I think that is a key differences in these wars.

* with a notable exception of the PaaS offered by Salesforce.com. While I think force.com will do well, I don’t see it winning the Cloud Framework war.

Peter

Hey William – enjoyed this post a lot. Some good food for thought. I would say VMware’s acquisition of SpringSource is an artifact of the app server wars – indeed it might be seen as heralding the last chapter – rather than the ongoing cloud framework build-out.

The winners in any technology wave are the best packagers- whether that’s Sun packaging Unix workstations, Dell packaging Windows and Intel, Compaq in the PC Wave, Apple with the iPhone, IBM with System/360 and so on. The race is on, as you say, to package Cloud frameworks. Someone is going to win very big indeed.

May the best packager win.

Yep, App servers and Cloud Frameworks will merge. Grabbing the rifle and preparing for a long battle ;o)

An interesting item on wmw:
http://www.redmonk.com/jgovernor/2010/04/21/vmwares-springsource-redis-and-rabbit-acquisitions-a-database-play-is-emerging/

“Developing cloud based apps will require new data management and storage models. VMware is getting well ahead of the curve”
“As alternative data stores become natural targets for new application workloads VMware does indeed plan to become a database player, or NoSQL player, or data store, or whatever you want to call it.“

vmw is morphing into a very different company!