Recently, I was asked to comment on a large institution that has specified Cisco in an RFP for networking and telephony gear. I got the call because there had been what appeared to be a similar instance with my own city, San Jose, which had sole sourced to Cisco. The award had been seen as embarrassingly improper. This then dovetailed with something fellow IT Business Edge blogger Dennis Byron's post, "SMEs: The Right and Wrong About a Short List of One."
Sole sourcing almost always looks like some kind of nefarious insider deal and, in my experience as an auditor, it almost always triggers an investigation that assumes guilt rather than innocence. It turned out that in the second instance neither the institution nor Cisco had really done anything wrong, but because of the way the RFP was written, it set off all kinds of alarms with the local press, which could have done damage to both entities' images.
When It Works
In his piece, Dennis points to Microsoft as one of the technology providers that is often sole sourced. There are likely instances where this is either well justified or a bad thing. In fact, it could be both in the same organization, depending on which technology we are talking about and what the intended use is.
One of the first measures of which technology or product should be selected is the skill set of those implementing it. Take a Windows offering and put it into a Linux or UNIX shop and it probably will underperform expectations significantly -- unless the skill set in that shop and the affinity for the platform is changed out as well. Even then, you'll have the problem of people not knowing the business and the costs associated with employee displacement and learning on the job, most of which are not captured but clearly adversely affect the success of the company.
Conversely, take a Linux or UNIX product and drop it into a Windows shop and you'll have a similar problem. People do not like to retrain and most, in my experience, don't. My belief is that much of the animosity that exists between Linux/UNIX and Microsoft initially resulted from failure to use the "embrace and extend" strategy that was so successful with Office. Maslow teaches that folks will do incredible things if their livelihood is put at risk. Having a customer base help create a competitor like Linux is unprecedented in any market and clearly meets the definition of incredible.
In short, the only way that a Windows product will be better than a Linux/UNIX product in a UNIX shop would be if it were generations better technically and came with a staffing change, including outsourcing, that could support it. The same would be true going the other way, creating an incredible barrier for both sides.
So I do believe you can, with some justification, allow the platform to follow the existing skill set. But can you also justify going single vendor for other things?
Cisco, HP, IBM and Microsoft
There is an inherent advantage that goes with sole sourcing a vendor -- this vendor owns the solution. Cisco, HP, IBM and Microsoft are broad-spectrum vendors that often are sole sourced in the companies, large or small, that use them for a variety of solutions. IBM's Global Services organization has historically given it the advantage here, recently challenged through HP's aquisition of EDS.
Having worked with and in both shops that aggressively bid everything and those that relied exclusively on a single vendor, I can say categorically, when asked which is preferable: "it depends."
Much as it is with the platform decision above, there is a skill set issue, but there are two additional aspects to consider: how tightly the solution integrates across the sole sourced vendor's offerings and who owns the solution that results.
Highly mixed shops seem to have the most vendor problems, primarily because when a problem results, those that have provided the components point to each other as the source of the problem. Where there is a firm like EDS or some other service provider playing the role of general contractor, this can be somewhat mitigated.
Going to a single vendor for a complex solution assures that people focus on the problem rather than who owns it. The vendor is not motivated to disparage its own products, giving a stronger impression that the problems will actually be resolved. Back when I was doing surveys, the happiest companies -- as measured by how satisfied they were with their primary vendors, regardless of whether they were all HP, all IBM, Oracle, or all Microsoft (I expect the same would hold true of Cisco or EMC) -- were those that sole sourced. These firms simply did not have the ongoing aggravation that mixed shops experienced.
However, if the primary vendor goes south, as IBM did in the 80s, it can be catastrophic because the dependencies are so deep. In addition, the vendor knows it owns you, and getting competitive pricing in the out years becomes problematic and creates increasing audit risks.
I'd Still Prefer Sole Source
I don't care for the ongoing aggravation that a mixed shop creates. To me, eliminating that aggravation is worth the reasonable, and I stress the word reasonable, risks typically associated with selecting a single-vendor solution. In the example that started this, once I understood what the institution was trying to do, I agreed that its selection of Cisco was both proper and would likely result in some very real cost and operational benefits. The CYA aspect, however, needed some work.