Monday, January 01, 2007
Thoughts on Open Source Franchising
Let's analyze the posting:
Roberto cites IDC and Forrester reports that indicate that the vast majority of enterprises want basic IT services (support, installation, maintenance). Even so, IDC found that 38% of enterprises hire outside consultants to do their migration services and 39% hire outside consultants to do implementation services. For these same services, 59% and 66%, respectively, do the work themselves, internally.
This means that either a) this sort of work is best/cheapest done internally or b) no one has created a viable model to do these basic services better/cheaper/faster.
There are several problems here that of course start with metrics supplied by large analyst firms as they are too freakin high-level to be meaningful. First, the notion of hiring outside consultants is driven by a majority of factors including the ability to get someone who has done it before which increases success, sometimes it is more cost effective where cost is about speed to market not salary vs consulting rate, the ability to offload perception with things go horribly wrong. You will never see metrics on the last statement but yet we all know that it not only happens but is pervasively done at a rate a lot higher than 38%.
Matt also left out a choice which I think is especially crucial in open source which is the fact that the vast majority of enterprises want basic IT services that integrate easily out of the box which points to an age-old problem of figuring out how to get vendors to work with each other. Since we can't get interoperability, we settle for integration.
Consider the ECM space in which Matt plays. Let's say I want to integrate Alfresco as my ECM product with any BPM product. Vendors will evangelize standards such as JSR-170 but will not speak about the fact that in terms of integration, I need a hell of a lot more standards than just this. How do these two integrated things share identity? How do these two integrated things share authorization models? How do these two things not create overlapping metadata that doesn't require syncronization and reconsiliation? How do these two things collectively utilize a common resource such as a portlet?
The conversation needs to move away from abstract high-level statistics collected by industry analyst firms towards conversations that talk about interoperable architectures across vendors. Don't tell me about your wonderful stack, I want to know how to build with Legos.
Likewise, I would say that the franchising model has several perspectives that are inaccurate at best. In his table, he lists that us customers demand for buying services could be to offer services at fixed price. Do you know why enterprises like fixed pricing? It is not for the reason you think...
Fixed pricing exists to avoid the pain points of licensing software! Do you know how much time is spent inventorying stuff within the average large enterprise just to comply with licensing terms? Does human effort expended on this task provide a business return? Paying a fixed fee is the only way to make this problem go away. Enterprise licensing from the vendors perspective tends to mean that an enterprise can use as much as they want while the enterprise perspective unfortunately may or may not be about usage but definetely is about avoiding accounting.
Consider the fact that I have been noodling introducing secure coding disciplines to my enterprise. I want to go and purchase tools such as Ounce Labs, Fortify Software, LogLogic and so on. I am a firm believer in having a consistent desktop where everyone simply gets the same image which aids in reducing desktop support costs and increases the potential for success in a disaster recovery scenario. I know that at any one time, that at best this tool will only be used by two or three developers concurrently. How do you think the vendor will require me to license their software? Will it be based on utilization, the value it brings to my enterprise or on something that they can account for?
Note that the above scenario is not just a problem with desktop oriented software but also exists in the data center as well. Consider which closed source J2EE vendors tend to lock their licenses based on IP address. Now consider the desire of the enterprise to use all those damn servers running at 10% utilization and their need to enable grid computing and virtualization which both cause workloads to move at will. Sure, a vendor can be very customer-service oriented and issue new license keys quickly but that simply doesn't work in a dynamic environment.
The one comment that does make sense is that vendors should participate in tech clubs to drive standards. For the latter problem, the enterprise can express a preference for open source but this is only a half-answer. Maybe, at an upcoming OSBC session, we could figure out how to ask attending vendors to start creating meaningful standards in the ECM space? Maybe even you could tell me what it would take for Alfresco to start the conversation with Interwoven, Documentum, Filenet and OpenText to create meaningful standards in the ECM space to enable interoperability?
The other statement that is worthy of exploration is Reducing software production costs is feasible if there are large amounts of volunteers and how enterprises can participate. The funny thing is that I am a known participant in the open source community and yet only a single vendor has asked me to contribute! What if while monitoring the downloads of open source software for leads, the sales guy on the other end also asked if us enterprisey folks had a desire to contribute. The discussion shouldn't end simply because I didn't buy something as there is a lot of value being left on the table without forgeting that value comes in many forms...
Links to this post: