...and another
thing

Microsoft's current server stack gives Jon Honeyball a Roy Sheider moment.

You recall the famous shot in Jaws. It’s on the beach and the police chief, played by Roy Scheider, has just realised that there is a huge person-chomping monster in the water. To give gravitas to that sinking feeling moment, director Steven Spielberg came up with a completely new sort of camera shot. As a result, Chief Martin Brody’s face stays firmly in the middle of the shot while the background does a stomach-churning zoomy contortion.

I got that same feeling a few weeks ago. I was visiting one of my corporate clients where I do some ongoing consulting. This firm is still happily sat on the Server 2003 family platform,. Everyone knows how it works, there is a fully-sorted disaster recovery plan, and the whole lot trundles along with a reliability that verges on the terminally boring.

The client has been mumbling about moving to Server 2008 along with Exchange Server 2007 and the newly escaped (or ‘finally released 6 months overdue’) SQL Server 2008.

To be frank, I have been resisting the move, not finding much there to really give this particular organisation the sort of business value to justify the cost, hassle, retraining and general opportunities for screw-ups and upsets. But the IT Director was firm – he wanted to talk about upgrading.

So we sat down around the boardroom table, and started to cost things out. To make sure we hadn’t missed anything, I fired up a Web browser and went to the Microsoft Server Web page. Here I found a list of the current applications which Microsoft has on the server platform. It runs as follows: .NET Framework, Antigen, BizTalk Server, Commerce Server, Essential Business Server, Exchange Server, Expression, Forefront Family, Forefront Client Security, Forefront Edge Security, Forefront Management Console, Forefront Security for Exchange, Forefront Security for SharePoint, Forefront Stirling, Forms Server, Groove Server, Host Integration Server, Identity Lifecycle Manager, Office Communications Server, Project Server, Search Server, SharePoint Server, Small Business Center, Speech Server, SQL Server, System Center Family, System Center Configuration Manager, System Center Data Protection Manager, System Center Essentials, System Center Mobile Device Manager, System Center Operations Manager, System Center Virtual Machine Manager, Virtual Server, Visual Studio, Visual Studio Team System, Windows High Performance Computing, Windows Server, Windows Server Update Services and Windows Storage Server.

At this point, I had my Jaws moment. Of course, we wouldn’t need to use all of these technologies. A subset would do. But once we started on a tick-list of ‘nice to haves’, the list was worryingly long.

Then I decided to be unhelpful. I pointed out that ‘Windows Server’ existed on the list as just one item. And that ‘Server’ actually covered a huge area of technologies, such as Active Directory, DNS, DHCP, IIS and at least another page-full of acronyms. The IT Director and I looked at each other and we both had another Jaws moment, because we both came to the same rather shocking realisation at the same time.

You see, back in the days when Microsoft had something called BackOffice, you could be a ‘generalist’ on the entire platform and be a specialist in one area – Exchange Server, for example, or Terminal Services. So we had two layers in our IT teams – the generalists who could keep an overview, and the specialists. Those specialists would look after the minutiae of the big packages, and one person could do a decent enough job if properly trained.

Today, after our gut-wrenching moment, we realised that Microsoft had slipped in a third level. Yes, we still have our generalists although their technical knowledge of specific solutions is weaker than ever before. Hopefully they can now fall back on those irreplaceable items known as ‘experience’ and ‘common sense’.

Then we have a new layer: a generalist in each of the main areas. So you could be a generalist in Exchange Server 2007, but probably not a real specialist in the whole thing. It’s just too big – do you really know all the pan-site design stuff and also the IP telephony? What’s the best storage architecture and can we have some OWA customisations too? No, today you can know much of most of it, which makes you a generalist in Exchange Server.

The same is true in SQL Server. Again, it has so much in it that you cannot know it all. Multi-site replication design? Business Intelligence and data mining? Query optimising and best practises for the indexer? Take Sharepoint Server: how many people know how best to design and deploy, and can also develop an app? Or customise the entire stack? Or handle clustering and load balancing? They can’t: they can do some of it, but not all.

So now we have a new breed of specialists at the third level.

Apple iPhone AppStore

As predicted by anyone with half a brain, the iPhone 3G has been a huge success. Far more important than this, however, is the Rev 2 firmware which has been made available to users of all versions of iPhone. With this comes the new AppStore which has been an incredible success. In a few short weeks there were over 1,600 apps on the store, many of them free. But over thirty million dollars was collected for paid-for software in the first month, and a few vendors have made a lot of money indeed.

Compare and contrast to the Microsoft smartphone development support. Yes, there is a very good SDK, but then the wheels fall off the Microsoft platform. There is no coherent clear and obvious place to buy software. You can’t install over the air. There is no straightforward integrated payment system. There is no upgrade path.

Worse still, Microsoft is starting to leak that next year’s Windows Mobile 7 will be “incredible” and have “astonishing touch support”. Well, should I buy a Windows smartphone today safe in the knowledge that when all this goodness comes along, I can upgrade my firmware? No. Microsoft doesn’t offer end users an upgrade path for the OS – it’s up to the third party hardware vendor to decide whether to offer that or not. So there’s no reason whatsoever to buy something today, and so no incentive to buy apps for it today either.

After all, if you offer me jam tomorrow, I’ll wait for tomorrow. Or the day after, to allow for the inevitable service pack. If I have the cash today, I want jam today. Quite why Microsoft doesn’t understand this is completely beyond me. These are the hyper-specialists who actually understand only a subset of a product, but they know it well.

Once you actually realise this, and have your Jaws moment, you realise that your current staffing plans are toast. You need to start again, think carefully about how you are going to lift this off the ground and ensure that you have either enough skill-set in-house or access to it from outside. And that you have enough understanding to ensure that you don’t get into that house-building conundrum when the brickie says, “Oh, those pipes are nuffink to do with me, mate – talk to the plumber.”

And here is the really scary thing. Do your newly created middle-layer app-specific generalists actually know enough to be able to have these conversations with their super-specialists and also be able to present this information in a coherent, correct and reliable fashion to the first level generalists who are trying to get the building built?

There is no doubt that the Windows Server ‘family’ for 2008/09 is staggeringly powerful, but the staffing needs have been left behind. And that is what will keep the deployment of truly best-of-breed full-reach solutions limited to only the bravest or most highly funded of sites. I don’t know how Microsoft can help to fix this, but they need to fix it now.

Ozzie’s last chance

Just a quick note on the forthcoming PDC – the Professional Developers Conference to be held in Los Angeles in late October. PDCs are very special things – they are the point at which the arm waving and bullshit has to stop, and the real delivery has to start. The target audience is the huge third-party development community, both standalone and corporate-hosted. We go to be wowed, but also to see a clear credible vision for delivery of new technologies in our applications. This PDC has already been delayed by one year, and Microsoft has been indulging in much arm waving and airy-fairy nonsense around ‘Software As A Service’ while the real deliverables have been increasingly noticeable by their absence. The PDC keynote is where Ray Ozzie has to deliver the goods. He will not get another chance.

...and another thing

Microsoft's current server stack gives Jon Honeyball a Roy Sheider moment.

Apple iPhone AppStore

Ozzie’s last chance

...and another
thing