(written couple of days ago on a plane – posted late – Blame it on the jet lag or all the iForum running around!) about 45000 miles across 5 different countries in Asia Pacific, I am on my way back to the US. Trust me, I had my share of cancelled, unconfirmed and delayed flights and long hours in airplanes, airports or hotels. But, I spoke with several IT professionals working at our customers, prospects, partners, competitors and industry analyst organizations.
If I were to summarize one observation and take-away, I think it would be around the confusion that exists in the market for the wide range of technologies that are now available for IT to leverage. It has been relatively simpler world for IT so far (at least when compared to what we have for us in the future) – you would install your desktop and applications on users devices, lock those devices down and you had your desktop infrastructure ready for most part. And then for apps that did not work in that environment for one reason or the other – either because your users required access from different locations, or because they were too hard to manage on desktops or they did not perform well on user desktops, you would decide to centralize them. That sums up the current state of architectures that you will see in most organizations. Application Lifecycle management ends up being an afterthought and is now being put into place as an incremental technology add to existing architectures.
However, there are now relevant market catalysts that are now forcing the organizations to re-think their architectures. The top 3 catalysts are:
a. Refresh cycles – The dreaded PC Refresh Cycle always gets you. There will be about 400M PC Refreshes that will occur between now and 2010. Ouch!
b. Compliance and Security – IT made some quick band-aids for solving their immediate problems with regulations. Now, they are turning at the strategic re-architectures for their sustained ability to follow regulations more effectively.
c. Growth – Believe it or not, this is acting to be one of the biggest catalysts – where IT are being asked to support organizational growth through M outsourcing, offshoring, etc. in a very agile manner – causing IT to think about new architectures that they can roll out broadly.
At the same time, there are myriad of technologies available with overlapping benefits, even though they all have a place for themselves. Example – streaming, virtualization such as CPS/TS, desktop streaming, Virtual Desktops or Dynamic Desktops as you will hear it from Citrix during iForum, etc.
I found several IT professionals confused about how they can pick the right architecture as they go through their architecture evaluations. what I see works, I will offer one recommendation that may sound simple and obvious, but is hardly followed and can actually be very hard to implement due to political boundaries within organizations.
My mantra is to STAGE DESIGN – i.e., separate your desktop delivery strategy from your application delivery strategy.
We often try to create an architecture that has desktops and applications all bundled into a single environment for the users. However, by separating the two, I have personally seen how IT organizations have been able to come up with a cost-effective and agile architecture. So, how do you build it?
1. Pick desktops for your users – split them into three categories – Office workers (those who spend a large majority of their time connected to the network) or Mobile Workers (such as Road Warriors, etc.). In most organizations, your office workers will constitute about 80% of your workforce. It makes sense to then pick the right desktop delivery strategy for these users. The desktop delivery strategy includes – devices, networks and as Citrix calls it Dynamic Desktops (delivered not deployed) – read more at http://www.citrixvirtualization.com. Different types of office users will need different levels of personalization – so you can build the right type of desktop for each user based on their needs. Dynamic Desktop is a broader topic for a post in the future. For users in mobile categories, you pick the appropriate devices (you may need more than one) and either decide to install or virtualize their desktops. Think about your applications – Now you profile your applications and pick the right architecture for delivering those apps. Some apps will be desktop based and some large line of business apps such as Oracle Financials or SAP will be best delivered via technologies like CPS. The more an app looks like client/server the more economies CPS will provide for the environment.
This could mean that you can have two tiers of centralization. One tier of published desktops and another tier of published applications. In fact, the most succesful implementations of CPS that I have seen are based on that architecture. This architecture provides the best modularity and efficiency for IT teams. Does it require extra hardware? At the end of the day, our customers tell us that it does not because they have optimized the hardware for the type of applications through silo environment and they are able to get the right hardware for the capacity and build out this two-tier environment with almost the same amount of hardware as they would have built with one-tier approach.
Bottom line – as you are thinking about new architectures, try to split your design into two distinct stages. The more decoupling you can do in both your design efforts as well as your architecture the better it will be for long term even if it requires purchasing a little bit of extra hardware to begin with. Those capital costs will very quickly offset the gains that you will get in your environment over time.
I am looking forward to get back to the US and head to iForum – few more days of hotel rooms and then I can go back to home!