I recently attended a presentation at which the CA (formerly Computer Associates) Directory of Strategic Alliances, Dan Woolley, spoke about how CA is supporting Windows Server 2008.
CA is not a company that I associate with bringing products to market quickly and I understand why companies are often reticent to invest in research and development in order to support new operating system releases. Hardware and software vendors want to see customer demand before they invest – just look at the debacle with Windows Vista driver support! There are those that blame Microsoft for a lack of device support in Windows Vista but they worked with vendors for years (literally) before product launch and even now, a full year later, many items of hardware and software have issues because they have not been updated. That’s not Microsoft’s fault but a lack or foresight from others.
It’s true that some minor releases are probably not worth the effort, but supporting a new version of Windows, or a new version of a major server product like Exchange Server or SQL Server should be a no-brainer. Shouldn’t it?
It’s the same with 64-bit driver support (although Microsoft is partly to blame there, as their own 64-bit products seem to lag behind the 32-bit counterparts – hopefully that will change with the "Longhorn" wave of products.
Dan Woolley’s presentation outlined the way that CA views new product releases and, whilst his view was that they are ready when the customers are ready, from my perspective it felt like a company justifying why they wait to provide new product versions.
CIOs expect infrastructure to be extensible, stable, flexible and predictable
He made the point that CIOs expect infrastructure to be extensible, stable, flexible and predictable (they abhor change as the impact of change on thousands of customers, users, and servers is difficult to understand) and how they:
- Deliver services to facilitate corporate business (so require a stable infrastructure platform).
- Work to maximise IT investments.
That may be true but Woolley didn’t consider the cost of running on legacy systems. Last year I was working with a major accounting firm that was desperate to move away from NT because the extended support costs were too high (let alone the security risks of running on an operating system for which no new patches are being developed). As recently as 2005, I worked with a major retailer whose back office systems in around 2000 outlets are running on Windows NT 3.51 and whose point of sale system depends on FoxPro for MS-DOS! Their view is that it works and that wholesale replacement of the infrastructure is expensive. The problem was that they were struggling to obtain spare parts for legacy hardware and modern hardware didn’t support the old software. They were literally running on borrowed time (and still are, to the best of my knowledge).
CA’s view is that, when it comes to product deployment, there are five types of organisation:
- Innovators – investigating new products in the period before it is launched – e.g. Microsoft’s Technology Adoption Programme (TAP) customers.
- Early adopters – who work with new products from the moment they are launched up to around about the 9 month point.
- General adoption – product deployment between 9 months and 4 years.
- Late adopters – deploying products towards the end of their mainstream support (these organisations are probably running Windows 2000 and are only now looking at a move to Windows Server 2003).
- Laggards – the type of customers that I described earlier.
Looking at the majority of that group, there are a number of deployment themes:
- Inquiring (pre-launch).
- Interest and testing (post-launch).
- Budgeting (~4 months after launch)
- Prototyping and plots (~1 year after launch)
- Deployment (~18 months after launch)
- Replacement/upgrade programmes (~5 years after launch – co-incidentally at the end of Microsoft’s mainstream support phase)
- Migration (7+ years after launch – onto another platform altogether).
What is interesting though is that there are also two distinct curves for product deployment:
- Sold licenses.
- Deployed enterprise licenses.
This is probably because of the way that project financing works. I know from bitter experience that it’s often better to buy what I need up front and deploy later because if I wait until the moment that I need some more licenses, the budget will not be forthcoming. It seems that I’m not alone in this.
CA view their primary market as the customers on a general/late adoption timescale. That sounds to me like a company trying to justify why it’s products are always late to market. Windows Server 2008 will launch early next year and serious partners need to be there with products to work with the new operating system right from the launch – innovators expect a few problems along the way but when I’m trying to convince customers to be early adopters I don’t want to be held back by non-existent management agents, backup clients, etc.
Windows Server 2008 is built on shared code with Windows Vista so the early hiccups and device adoption should already have been ironed out
CA’s view supports the "wait for service pack 1" mentality but then Woolley closed his presentation by stating that CA builds on Microsoft platforms because they consider them to be extensible, stable, flexible and predictable and because they will allow the delivery of service to facilitate corporate business imperatives and maximise IT investments. He stated that CA has been working with Microsoft on Windows Server 2008 architecture reviews, design previews, TAPs and logo testing but if they are truly supportive of Microsoft’s new server operating system, then why do they consider their primary market as not being ready for another year?
Once upon a time hardware led software but in today’s environment business is supported by software and the hardware is just something on which to run the software. In today’s environment we have to consider a services model. Microsoft’s move towards regular monthly patches supports this (they are certainly less focused on service packs with the last service pack for Windows XP – the client operating system with the largest installed base – having shipped over three years ago).
Windows Server 2008 is built on shared code with Windows Vista so the early hiccups and device adoption should already have been ironed out. That means that Windows Server 2008 should not be viewed as "too new", "too disruptive" (it will actually ship at service pack 1 level) and, all being well, the adoption curve may be quicker than some think.