James Governor's Monkchips

Why do we fetishize “best practices” but not “best infrastructures”?

Share via Twitter Share via Facebook Share via Linkedin Share via Reddit

Reinventing Local Government ォ Manage By Walking Around

Jonathan Becher, CEO at Pilot Software, is a contrarian thinker in corporate performance management, and he asks a great question in his latest blog entry, jumping off a speech by Bob Hanson, CIO of Sarasota County Florida.

“Sarasota County spends more than $100K every year to operate their email system.

Since Sarasota County is probably larger than most, assume the average yearly cost is $25K

Therefore, something like $2.5B is spent every year on municipal email systems.

What if we replaced the 10K individual systems with one shared email infrastructure modeled on Yahoo! Mail or Microsoft Hotmail? It’s certainly technically feasible and we’d end up with a system that would cost a fraction of the $2.5B to operate but that would have greater redundancy, reliability, and security than most of the organizations enjoy today.”

Becher talks to the goodness of best practices but it seems to me the language of best practices is often utterly bankrupt. A best practice, to my mind, must be based on direct practitioner experience. Unfortunately in our field it can mean almost anything.

In the web services community, for example, “best practice” is often used as shorthand for implementation of a standard.

The disease of calling arbitrary processes or implementations best practices holds the industry back, again and again.

In this age of shared services, shouldnt we be looking for “best infrastructure” though, rather than best practices? After all, the best infrastructure has already proven itself in production, while best practices on the other hand have often not proven themselves anywhere other than on a technical document.

I don’t like to fetishize the services provided by web companies either- but hotmail et al obviously do scale for the task in hand.

What companies would benefit most from “best infrastructure” thinking? Best in class service providers. EBay, Amazon, Google, Fedex and so on. And Sun – CEO Jonathan Schwartz is puttiing most of his eggs into the shared service provider basket. Jonathan wants to focus on customers that potentially provide “best infrastructures” or “best services” , and is prepared to focus less on those that are obsessed with building to “industry best practices”, long after the market has established them. Sun is prepared to bet on best infrastructure. Are you?

Who would suffer in a best infrastructure world? Likely IBM for one. IBM Global Services and Business Consulting make tens of billions of dollars selling best practices. Even SAP, which sells best practices instantiated in code, is still selling best practices, rather than the best solution to the job. IBM needs to become a switch in the sky if it is to play more effectively in a best infrastructure world. Currently its customers have too many opportunities to poorly implement its software (the same is obviously true of Microsoft .NET)

I am talking about the difference between specification and implementation.

Of course political issues make the adoption of best infrastructure difficult, in any sector, from local public sector to the biggest enterprise. Over the last thirty years organisations have concentrated efforts on managing their own assets, and “IT competitive advantage”. But what has doing so bought? In many cases large management overheads. But everyone fights for budget, whether they need it or not. Ever spent money in order to get a budget for next year, rather than because you need to spend the money for new infrastructure now? Exactly.

If we buy into ITIL say, a best practice framework for IT Service Management, that means a significant investment in training and cultural change. We have to a create a culture of excellence, and best practice. In some cases it would surely make sense to just subscribe to existing successful management services rather than building our own. That’s one of the problems with trying to sell ITIL to SMBs. They might be better off with a simple managed service.

Best infrastructure can also be fetishised in its own right. The UK government is currently spending upwards of ten billion dollars on a project that takes purchasing power out of the hands of local medical practices, and hands it too five unelected “regional service providers.” Of course, none of these service providers has ever actually provided the service in question, before so to my mind they are neither best infrastructure nor best practice. Needless to say the players are missing deadlines and pointing fingers.

As a taxpayer I would like to see the government recommend choices from service providers that can show a working system for the service they plan to provide. I am not saying the organisation has to already be in that exact sector, but at least be solving the problem in question. Thus for example, local government organisations should be encouraged to try salesforce.com – because it offers a working system, rather than some best practice ideas and piece parts to implement them.

I haven’t used the SOA word yet in this piece, but that is a likely context for a discussion of best infrastructure going forward.

Of course the world isn’t going to move wholesale to a try before you buy, acquire service rather than products world, whatever salesforce.com would like. But when you see CIOs from public sector organisations start questioning whether they have too much budget and too much freedom to implement stuff, you have to wonder whether the shared service bug isn’t catching.

So lets fast-forward to a world of bureau-based computing… everything old is new again.

disclaimers: Microsoft and IBM are clients. Pilot and salesforce.com aren’t.
tags: , , , , , , , , , , , ,


  1. Even the expression “best practice” assumes that it can’t be done better – which is patently false. “Best practice” is what worked for SOMEone else, SOMEwhere else, SOME other time – the chance that it will work as well for YOU, NOW, in YOUR place is extremely remote.

  2. James,

    Great post. Some interesting thoughts. I like how you are abstracting it out. My additional thoughts are around the fact that software, hardware, infrastructure and services are usually what are used to build a solution to a business process, function, etc.

    In a shared services or in any organization you should be looking to leverage the best solution to solve that problem. In web 2.0 that is often the service itself and you rely upon the provider to be leveraging best in class components.

    I think that this is one of the reason that SaaS has some legs going into the future as today’s effort at being a best in class service leveraging the best in class components underneath without the customer paying for the mashup of these components from scratch.

    Your email example is a perfect one.

  3. James,
    I’ve never liked the term best practice either, even though much of my day job is spent driving what we call here in enterpriseyville, the best practice network. (If you have a better idea for a name let me know)

    For most of what companies do emulating others makes a whole lot of sense. The skill of copying is very undervalued. The trick is to know when to copy and when to innovate.

    Last week I listened to a major reinsurance company explain how they managed their SOX project. For the other companies in the room, this exposure to “best” practice, or “a damn good” practice has significant value. So copy it, and spend your “innovation quotient” elsewhere.

    Best infrastructure is part of the story. If the infrastructure isnt upto the job,then the process is damned from the outset.

  4. Too often best practice is defined by those with a vested interest in flogging a service – aka Gartner, McKinsey et al. What I don’t understand is why more compnaies don’t just come out the woodwork and say ‘we did this or that’ and leave others to decide if it might be an *approach* that would work for them.

    The reality is there is no such thing as competitive advantage in ‘what we did’ because even assuming I get it – what are the chances of me replicating that. I’m not Dolly the Frigging Sheep and neither are businesses.

  5. It is hard to disagree that SMBs are better of getting services from a solid service provider. Providing “best infrastructure/service” is difficult and expensive. Economies of scale does help.
    As stated in your example, as a (very) small company we are using Yahoo as our email provider. Can we operate our own email server? Sure, but to have highly available, secure email service would take significant amount of time/resources.
    For larger organizations like Saratoga county in your example, the decision is not as clear cut.

    When using a service from a massive service provider like Yahoo/Google/Microsoft, there are significant challenges. These organizations may have the “best infrastructure”, but they do NOT have the “best service (support)”. They figured out how to scale their infrastructure to serve millions but they do have not figured out (at least yet) how to scale their service, which turns out to be more difficult, and resource intensive.
    If you work with a massive service provider, things look great when everything is working, but when something goes wrong, your options are limited. There is typically no SLA, and support is minimal. It is hard to get someone to talk to and problem resolution times can be unacceptable.
    Even as a small company, when we have experienced problems with email, it’s a big deal! We’re dependent on it. And as a customer, you’ve much less leverage, as the revenue from your business is a drop in the ocean for these providers.

    I am also not sure that the best infrastructure is a replacement for best practices. If you use services from a best infrastructure providers, you may no longer need best practices for providing these services, but you’ll need new ones. What should be the criteria for service provider selection? How do you measure the quality of the service? How do you diagnose/locate problems? Where does the problem originate from, your network? which service provider (application, WAN)?
    The challenges organizations will face when using services from different service providers in conjunction with inhouse services are significant to say the least. So we need best practices to deal with them 🙂 I want to know how others deal with the problems I’m facing with, what the best way to handle them, what are the gotchas. Best practices is about learning from the experiences of others instead of repeating them.

  6. OK, I may have been quick to use the phrase “best practice” and in retrospect, that’s not at all what I meant. Bob Hanson did a great job talking about the need for shared infrastructure and I completely support him on this idea. For my corporate job we use a CRM program that’s available as SaaS — that’s so we can reduce our support cost through sharing the infrastructure with other customers. It’s analogous to a timeshare for vacation homes.

    My point is that the public sector — especially local gov — can go a step further. They aren’t in competition with each other so they can share the information that they put into the shared infrastructure. By also sharing information, they can look for trends, understand what’s working and not working, and jointly come up with new practices. Because this is now a collaborative effort, these practices are likely to be better than what they would do by themselves. So, they may not be best practices but perhaps we could call them “collaborative practices”.

    Sharing infrastructure is a great advancement. It’s just not everything that can be done.

Leave a Reply

Your email address will not be published. Required fields are marked *