There’s been an interesting thread on the “Open Source Business’ Glass Ceiling” over the last couple of months that I’ve been meaning to comment on.  I think Marten Mickos’ initial categorization of customer willingness to pay for support is perhaps a little too simplistic; or maybe too abstract.

As a recap - Marten asserts that there are two types of organization

“… organizations that have more time than money and organizations that have more money than time.”

The former (A) are willing and able to spend time to become technology self-sufficient so they can save money on support contracts; the later (B) don’t have the time to build up the expertise (or the motivation) so instead are willing to pay someone for a support contract to save time.

Savvio - refines the model a little and suggests there’s a third category (C) of organization that “… has more money than time but is used to getting what they need for free and is comfortable enough with OSS to rely on their own skills”.

Let’s step back a bit. In my experience gained over the last 10 years working for both proprietary and open source vendors - I’ve met with literally hundreds (maybe thousands) of customers and rarely have I identified an organization that would be easy to pigeon-hole into Category A or Category B (or C for that matter). What I have seen are many cases where an organization exhibits a Category A or Category B behaviour (and often both). The only place where I’ve seen a specific decision to self-support were in two extreme cases - the first - web startups who’d decided they would keep costs low and use only free (0$) software and self-support (so this was driven by less money, not more time); the second - a huge e-commerce mega-site - decided to self support Apache HTTP Server and Tomcat because a) the risk was low (ie. very stable technologies with large user base) and b) the org. had created their own branch to meet some very specific requirements - so they had gained the expertise through necessity.  Nearly everyone else I’ve spoken to is either conditioned to buy support (and just does it as a matter of routine whenever they buy a product) or have specifically identified cases where self-support (or no support) is an acceptable risk - but I’ve very rarely seen this decision made across the whole organization.

With this insight I prefer to look at the situation this way :

Importantly it’s generally at the granularity of projects (because risk is different for different projects) and it’s dynamic - ie. an organizations willingness to self-support often changes. For example a web-startup may initially choose to self-support to maintain zero / low costs; but as the application becomes more valuable to the company (ie. as they recruit real paying customers) and as they focus more on business and less on IT - the project moves up the “Support Value Gradient” (Blue).

What’s interesting is the yellow area - ie. the “Self Support Value Gradient”. In the proprietary software model - this is very small - I would argue that the ability to self support on a product  a) that has no community and; b) for which the source isn’t available is extremely limited. In reality, “No Support” is likely to be the only option. This area is largest when the technology under consideration is a) very stable; b) in widespread use; to the degree that ’self support’ may not require any investment at all. An example, and I choose it because it’s vendor neutral is the Apache HTTP Server - it’s in widespread use and it is pretty stable. Of course - even in the case of Apache HTTP Server - as soon as you start including the various plugins, containers and tools to make it useful - you’ve now got a reasonably complex integration to maintain - depending on the profile of the deployment - the value of vendor support could still be pretty high.

At JBoss - we have a lot of ‘customers’ who are using JBoss technologies without paying Red Hat or anyone else a penny and I’m absolutely OK with that because when they do move up the “Support Value Gradient” - ie. when they see sufficient value in the use of the products and value in the services we provide - it will be a simple transition.

This is all just my opinion, based on my experience; at the end of the day commercial open source is still pretty young - so we’re all blind - trying to figure out the nature of the elephant. I also don’t think these concerns are limited to Open Source - they’re just more acute because customers / users have a choice and choice is a good thing.

Tags: JBoss, open source, Red Hat