Late last year, I wrote a paper on the challenges facing enterprise IT when they consider SaaS/S+S consumption. The full paper has now been published in The Architect Journal and is online at MSDN.

I summarise my points below, go take a look at the full paper to see why I draw these conclusions;

Consuming an external service reduces the operations issues you will face but introduces new challenges such as integration, contractual negotiation and reduced flexibility.

Impact ripples beyond IT: users are often given another set of credntials to access the service resulting in more complex provisioning/de-provisioning processes, the need to update helpdesk procedures. You also have to manage user training

You may be facing an EAI project: does the data to be held in the service need to be used by internal applications? Does the service provide the ETL capabilities your users need?

There may be infrastructure implications: do firewall rules need to be changed, are there certificate implications (e.g. smart-card authentication)

Operations will be impacted: as a minimum, you need to give your helpdesk team the new application escalation details however most organisations will need to perform some troubleshooting on access issues (e.g. is company firewall causing the problem, is it some change they've made to the client/browser etc) before a problem is escalated to the service provider.

There may need to be some deployment: even if the application is accessed through a browser, there are will be a minimum version required. It may be neeccessary to add the application domain to the Trusted Sites list, or deploy a JVM/ActiveX control which your users do not have rights to do.

Evaluate the legal implications: if your business is regulated, you need to work out if any regulations affect the data or service, and if so, how do you prove compliance? If you're not regulated, you still need to consider national laws e.g. data privacy.

In conclusion:

• Consumption of an external service by an Enterprise has implications beyond "selection and subscription". As an absolute minimum the contract and any SLAs need to be evaluated, more likely you're facing some form of integration project.
• SaaS/S+S as a delivery model is not yet mature enough to have a good enterprise integration story. Many SaaS/S+S apps are aimed at consumers or small businesses who have no integration requirements. Vertical soluitons fare much better but most still have a long way to go.
• The more important the application to the business, the more integration and due diligence is needed. The converse is also true.

This is just a summary, I encourage you to read the whole thing. Enjoy the read.

Kevin

I read the press release when news broke about the aQuantive acquisition last Friday. Like many press releases, if left a few questions in my mind. Fortunately there were some talking points in my inbox this morning which help clarify the objectives of the acquisition;

The deal is an extension of Microsoft’s big bet on the online advertising space.  With a total value of over $500 billion, $40 billion of which is online, there is tremendous opportunity to not only drive future growth for the company, but also to provide this industry with tools and services that will help advertisers generate the highest possible return on their ad investments and will help publishers maximize their revenue.

Our merger with aQuantive will enable the new team to strengthen our relationships with advertisers, agencies and publishers by enhancing our current advertising platform with complementary technologies.  For the first time, we’ll be offering display advertising solutions for all ad agencies and publishers on any websites; providing a choice while, growing the addressable market for Microsoft products and services. 

With these new assets, we also increase our capabilities to build and support next generation advertising solutions and platforms such as cross media planning, video-on-demand and IPTV.

Interesting stuff.

Kevin

On Tuesday 28th I delivered the 1st of my EMEA Virtual Server events, which will be touring to >25 countries across the EMEA region over the next 6 months. Sharing the presenting was Shai Ofek (Microsoft Corp, from Remdond) and David Miller (Avanade). I've posted the session titles and abstracts below, and linked to the PPT decks.

Microsoft Virtualisation: Today and Tomorrow

Virtual PC and Virtual Server have gained significant adoption in development and test environments, and are an important stepping stone to realising Microsofts vision of self-managing dynamic systems. Virtualisation technology is gaining a foothold in the data centre, enabling more efficient utilisation of hardware and helping IT react more quickly to changing demands. This session introduces the Microsoft Virtual Server technology, its capabilities, how its being deployed today and the benefits and challenges customers are seeing. We look at how Microsoft will evolve the technology with the introduction of Hypervisor in codename Longhorn Server. We’ll look at the steps Microsoft is taking to help customers and partners use Virtual Server effectively in heterogeneous environments.  Finally, we’ll predict how virtualisation will change IT, from the data centre to the branch, servers to the desktop.

Managing a Mixed Virtual/Physical Environment: Tools and Techniques

Microsoft believes the management of physical and virtual machines should be largely transparent. The same tools and techniques should be used to deploy, monitor, maintain, and backup physical and virtual machines. For this to happen, tools need to understand the difference between physical and virtual, and automatically adjust the actions they undertake to accomplish the same task for physical and virtual. This session looks at System Centre Virtual Machine Manager (VMM), Microsofts first product which bridges the gap between physical and virtual. We’ll look at the complete lifecycle, from provisioning, to backup, migration (physical to virtual and back) to monitoring, and how this fits into the management of the rest of the server infrastructure.  

Microsoft Virtualisation Deep Dive

What is the Architecture of Virtual Server, and how will this change with the introduction of Hypervisor? How does host clustering work, what are enlightenments? This session starts with a look at the components which make up Virtual Server, covers the changes introduced with Windows Virtualisation (Hypervisor), and takes a detailed look how the virtualisation stack works.  We cover the High Availability capabilities inherent in Virtual Server, investigate the benefits of AMD-V and Intel VT hardware virtualisation for non-Windows guests and cover some of the standards work Microsoft is pushing to improve the performance of disk and network for the virtualisation industry.

Avanade: Virtual Server notes from the field

If you believe the hype, you could be forgiven for thinking Virtualisation can solve most IT problems. In this session, Avanade will share the experience they have gained helping customers adopt, deploy and manage Virtual Server. We’ll look at typical workloads, why they’ve been virtualised, the challenges they needed to overcome to gain business benefits. And, of course what the business benefits were. Things like backup and recovery, load balancing and capacity management are all covered. In effect, the good, the bad and the ugly.

As usual, I welcome feedback.

Last Update: 09/11/06 16:18 GMT 

With any new software, the user has to go through a learning curve. When the software UI has changed significantly from a previou version its easy to get frustrated as you can't find the features you're used to, resulting in lost time hunting them down. Is training or self discovery the answer? Is it both? Is the answer something I've not thought of?

Over the years I've worked on many projects where the limiting factor in rollout velocity is clasasroom-style user training. The technical solution to migration/deployment is often highly automated and could migrate 100s of users/desktops per day, if only training had the same capacitty. I've often wondered if this classroom style training was really needed. It costs a lot of time and money, lengthens the project and the results are never validated; how many projects migrate two pilot groups, one who receive training and one who don't, then compare the number of helpdesk calls, user productivity and user experience between them?

Microsoft IT (the internal IT department) take the "do it yourself" approach: users can visit various intranet sites to self learn if they want, there are some regional classroom style training courses available, but rearely for beta software. As far as I know, few people take advantage of either. Granted, Microsoft are a company of largely tech-savvy users who should pick things up more quickly than some others, however most people I know feel training gets in the way of doing the job, and I know many many business people outside the IT department in customers feel the same. So, why do we continue to force business people onto training courses as part of a rollout? Perhaps we should let staff choose whether to take training?

Having installed Office 2007 on my XP laptop 8 weeks ago, and then migrated to Vista a few weeks later, I've gone through (to be frank, I'm still going through) a learning curve which is frustrating at times. Do I want training? No. Why? Because most training is generic, and will teach me stuff I already know or have discovered for myself. Other people will feel differently, so training should be available if users want it.

When Microsoft product groups go through product design, they put a lot of effort into the user experience (UX as they call it). The public are engaged in this process resulting in a range of user experience levels providing feedback. So, why do Microsoft change the UI of Windows in every release, and why did the Office UI change in Office 2007 so radically, when they know this will result in a steeper learning curve, and slower deployment? Simple answer: productivity. The changes come from studying how people use computers, then optimising the UI to speed up tasks.

I'd say that I can already do tasks more quickly in Office 2007 than in Office 2003, and day to day tasks on Vista (finding documents, browsing the web, working remotely and wirelessly in Starbucks) is about the same as XP, but system tasks (anything to do with the control panel) is still taking me longer compared to XP. Why? Because stuff has moved, which brings me to the point of this post: a list of things that have changed. Everyone is going to go through a similar "voyage of discovery" (others would call it pain), listing where stuff has gone might save you a few minutes.. I'll add to this post over time, feel free to add comments and I'll consolidate when I get a moment.

More to follow..

Kevin

Thinking about the challenges I've faced on projects over the last few years, I've noticed they broadly fall into three problem areas;

• Security
• Identity Management
• Operational Management

I doubt anyone is surprised by this list. Security is easily the most painful topic whilst its difficult to prioritise the other two. Initially, I thought the prioritised order would change throughout the lifecycle of a solution. Operational management isn't an issue during the design phase right? But when you think about it, you need to design operational management; you need to get the dev team instrument code appropriately, you need monitoring services, deployment services, configuration management services, reporting services etc. So whilst the activity related to each of the three areas above will vary across the lifecycle, I feel the areas are still the most prominent problems Infrastructure teams face. If you have a different view, drop a comment.

I find its quite useful to know these three areas are the biggest problems; you can focus extra effort and resources on them. You can look for best practices around them. For example, during design and implementation phases of my last project we set up teams broadly aligned with these problem areas and integrated them with teams focused on particular products. I think of these as horizontal teams focused on (primarily) non-functional areas and vertical teams aligned with functional areas, as shown in the diagram below (liked here, this image is stored on a public dogfood server, so it may not be available at all times).

Drop a comment on these trends; do you agree? Is something else causing you sleepless nights?

These trends help to focus my attention – is there a deficiency in a Microsoft product which is causing the issue, or is it a result of poor process, internal company politics or something else?

Between March and June '06, I had the pleasure to deliver an Infrastructure Architect Forum in many countries across Europe focused on the Microsoft Dynamic Systems Initiative (DSI).

DSI is not particularly well understood either across the industry, or among many Microsoft staff. The objective of the forum was to help customers and staff alike get a feel for DSI; the vision, where we're at in the roadmap and how it will affect products and employed by IT Operations teams.

Each Forum had a keynote presentation from one of the Partners at MWD Advisors, an analyst firm who focus on IT business alignment. MWD are somewhat unique not only because of their focus, but also because they provide their research for free.Their website is a great resource for subjects such Identity Management, Service Orientation and many others.

You can find their presentation, entitled "Rethinking Enterprise and Infrastructure Architecture" here.

The other three sessions were delivered at the Pakistan Developer Conference, detailed below. The only difference between these sessions and the Forum sessions is the PowerPoint template.

If you want to follow up on the Health Model concept or any of the large issues discussed in DSI, see here

Finally, I'd like to thank the many people I met in Portugal, Austria, Belgium, Italy, Poland and Pakistan for their insightful comments and excellent questions.

IT Forum is now only a few days away, and its success is now largely in the hands of the speakers and onsite logistics team. The success is measured by you; through completing evaluations.

Let me give you an insight into the importance we all - speakers, events team and ultimately the event budget holders - place on the feedback;

The event is measured inside Microsoft based on the feedback. If the overall event score shows year-on-year improvement we have achieved our goal.

We award the highest scoring speakers with a prize to encourage the best performance.

The speakers are highly competitive, in the speaker room there is a large plasma showing the current top speakers and their score - this is automatically refreshed every 60 seconds. Its quite normal for a speaker to check their score immediately after the session (in fact, its possible for them to check their feedback whilst delivering a session) - and most read all the verbatim comments. Speakers who score highly usually get invited back to the next event. You can work out what happens to those who under perform. Performance at events like IT Forum is often a discussion at the speakers annual review with their manager.

Top track prizes are awarded to encourage the best content selection by the track owners.

A low-scoring session will immediately get flagged to the relevant track owner and I. We find out the reason for the low score and take action accordingly, whilst the event is running. For example, sessions which finish early typically score badly – and rightly so.

There are other reasons some sessions under perform; misleading abstract, poor speaker knowledge, noisy or uncomfortable room. Some of these can only be captured and fixed if you provide verbatim feedback; please be specific. If you’re not happy with anything, we want to know.

During the event, I typically spend 30% of my time looking at the scores to see if we’re going in the right direction or if we need to make some changes. In the past, we’ve cancelled poor-performing speakers outstanding sessions so you don't attend a bad session.

There is one thing you can do to help make the event meet your needs; provide feedback which is actionable.

Some good examples of actionable feedback:

• Great session, the demo really helped show how to setup clustering
• The speaker needs presentation-skills training
• The room was too noisy

Conversely, here is feedback which isn’t actionable:

• Great session J
• Waste of time
• Get a better speaker

We want to know why the session was great, why it was a waste of time and how the speaker failed.

I’m confident the event will be a great success and you’ll find it valuable, with your feedback we can make next year even better.

Cheers & enjoy IT Forum 2005

Kevin

So here's the list of blogs I promised last week;

• Connected Systems Infrastructure, Sascha Corti: http://www.corti.com/weblogsascha/
• Core Infrastructure, John Howard: http://blogs.technet.com/jhoward
• Data Management, Gunther Beersaerts: http://blogs.msdn.com/guntherb/
• Identity and Access Management, Simon Veale: http://blogs.technet.com/veale
• Management and Operations, Olga Londer: None
• Messaging & Mobility, Simon Shepherd: None
• Microsoft Business Solutions, Lara Martini: None
• Office System, Mark Harrison: http://markharrison.co.uk/blog
• Security, Fred Baumhardt: None
• Windows Client, Jeff Wettlaufer: To be supplied

We've posted a partial list of the sessions we're plannning on the IT Forum website, and the list of pre-conference sessions has been updated. Both of these will get a further update next week.

Kevin