I had been meaning to write up OpenLogic for a while. These guys are smart and have a cool approach to managing open source distributions and interdependencies.
Anyway, Stephen got there first, which is good because it means I can focus on the language and context aspects of the OpenLogic value proposition, building on his analysis.
Some commentators criticize new terms like AJAX or Web 2.0 and treat language as some sort of snake oil. Technology is evidently expected to evolve, while language isn’t.
From my perspective though language evolves to create new contexts for innovation and process change and so on, in a fascinating interplay between technology and the spoken/written word. Kathy Sierra says we’re all marketers now, and I am with her.You are a marketer – so deal with it!
OpenLogic deserve some props. Not just they are doing the right things technically, but also from a language and t-shirt creation point of view. Jeff Nolan likes the company too.
So what did Stephen say?
BlueGlue can install any combination of components from a library of about 130 of the most popular open source projects (as well as sample applications & templates): Ant, Apache, Eclipse, MySQL, Tomcat, etc. While this feature might by itself be attractive to Windows users, who as discussed previously don’t have much in the way of native tools to manage non-Microsoft applications, Linux users are unlikely to be impressed solely by the ability to install applications because most have that already.
OpenLogic’s tool, however, does go beyond the abilities of the Linux package management tools I’m familiar with in a couple of areas. First, and most important from a developer perspective, is that the configuration, installation sequence [1], and post installation testing (pings and everything) is taken care of transparently. For example, let’s say I want to deploy an infrastructure for building a Java based web infrastructure: I can check off in the tool Spring, Struts, Hibernate and, say, MySQL and have them deployed and configured for me, with zero know how or intervention required on my part. It can even preconfigure applications in the library such as JBoss for external, commercial applications such as DB2 or Oracle.
Why is this important? Well, one of the criticisms of open source we hear frequently is that it’s too difficult to install, configure and manage on an ongoing basis.
Microsoft is also currently explicitly working to establish a frame that says OSS has a major problem with component independency. Jason Matusow calls this “the integration problem”. Microsoft pitches “integrated innovation” as an alternative to open source modularity (with its corrolary fragility).
One potential answer to that negative frame is to point out Microsoft’s own interdependency problem. That is, the user is not in control of component choices. Install one server and others are sure to follow. Install Exchange 2003, for example, and you also need to installed Active Directory; if you want Outlook Web Access you need to install IIS, and so on. Integrated innovation sometimes seems like an excuse for monolithic installations. Everything always seems to be connected to everything else. I need a media player for my OS ding?
But there is another possible answer to Microsoft, which is to solve the problem technically. SpikeSource and SourceLabs are both working on integrated tested stacks for deployment. Aduva has been in the space for a few years now. But back to OpenLogic. The firm’s BlueGlue platform is designed to offer the user control of the stack, and which components they choose to install. But in the world of open source particular components tend to run through rapid development iterations. How to keep up with those changes and manage the interdepencies from a production deployment standpoint?
As many of you will know Stephen is a big fan of Portage, which allows easy package management across Linux distributions. But BlueGlue is far more ambitious in scope.
OpenLogic is also playing with language to help explain what it does, which I think is a great approach. If you wanted to be able to automate and script the building and tearing down of a stack, with full interdependency management, why not call it a stack macro? Drink a couple more beers and hey presto stack macro becomes “stacro”.
A category is born.
The “stacro” idea plays directly into what I believe will be the sweetest of IT industry sweetspots, that is, the intersection of Software as a Service, Shared Source and Service Oriented Architecture.
SaaS – an annuity revenue stream, with ongoing updates as components are updated
Shared Source – i use the term so as not to deny Microsoft a place at the table. There is plenty of Microsoft code out there now being shared.
SOA – key to SOA is loose coupling, but loose coupling requires interdepency and component management. In the language of SOA we need a registry and repository to manage these elements.
All three of these industry trends converge with stacros. But then again, they also converge in the notion of automated configuration management databases, a growing and related market. I expect BMC, IBM and and HP to become stacro players going forward.
It would be great to see Microsoft establish a stacro strategy, too. In fact it may have to – most enterprises want to maintain control of the new components they introduce into their enterprises. Database security patching is a related example.
Of course users want to deploy to integrated suites of components. But they want to retain choice and competition of components. That is a promise of SOA. And SOA needs to work its way into the platform, not just the endpoints. If SOA is just glueing monolithic endpoints together we might as well just quit now and buy SAP.
Finally – you know what – even even if you hate the term stacro with a passion and nobody but me and Andy Grolnick ever uses it, it certainly helped my crystallise some thinking about some important issues that will need to be tackled before open source can show its true value. Talking of true value and the importance of language- BlueGlue – a potential IBM acquisition, anyone? It might be a nice backfill to Gluecode… speaking in language terms…
No Comments