Intel släppte i augusti sina två första 4-kärniga processorer för bärbara och världen tar ytterligare ett steg mot en hårdvaruarkitektur där program som inte kan utnyttja parallellexekvering får allt svårare att motivera varför. Inom en framtid kommer vi se 16, 32, 64, kanske 1000-tals kärnor i processorerna. Priset för en beräkningscykel kommer att gå mot golvet; men bara för dem som vet hur man utnyttjar arkitekturen (vilket är svårt).
I framtidens systemutveckling så blir följande ord relevanta:
1. Energi (hur ser jag till att inte slösa bort tomma exekveringscykler)
2. Utvecklarproduktivitet (hur bibehåller jag produktivitet när programmering blir 2-3 ggr mer komplex)
3. Tillförlitlighet i mjukvaran (speciellt om du har 1000-tals kärnor att debugga samtidig)
My colleague and friend Johan Lindfors took on a challenge this week at the Developer Summit Conference. Going under the name of the "Naked Coder" he challenged the audience to come up with at game idea and to implement that idea as a Xbox game using XNA in 24 hours....
Quite a challenge ! Here is the result
Finally the program for the coolest architecture conference this year is online at www.iasa.se
This Thursday I was participating in a panel at the ICT fair. The topic was Software As A Service and the panel had representatives from the user perspective, the legal perspective and a couple of SaaS vendors like Microsoft, Google, TrendMicro and Salesforce. We were discussing both benefits and disadvantages about going to SaaS.
Some highlights:
Lars Jerkland from Google highlighted the trend of moving consumer applications into the Enterprise, Kaja Narum from TrendMicro talked about security as as service; which basically is about introducing orthogonal infrastructure services in the Enterprise which is kind of an emerging market. Johan Zetterström from SalesForce talked about the cost and installation benefits of enterprise SaaS solutions. Jim Runsten, a lawyer at Bird&Bird talked about the importance of actually understanding what you buy and Alan Voreck talked from a user perspective about how SaaS applications for accounting firms seems to be a sweet spot.
I have always been very positive to SaaS solutions; they have a lot of capabilities that makes a lot of sense in many scenarios. However, when I was driving home in my car after the panel I realized that there is something about SaaS that makes me a little bit nervous about the near future. Let me elaborate:
Let's take a historical perspective. What has been the focus of IT projects in the enterprise for the last 5 years ? Most people would probably argue that is has been the change to a service oriented way of working and thinking. The purpose obviously to give your organization a more flexible, effective and agile IT architecture.
A lot of things has been done to support this vision: Single-Sign On Solutions, master data management, process externalization from legacy system, integrating these processes with the rest of the organization using standards based protocols etc... This has been possible largely by taking back the control of information and processes that were deeply hidden in old legacy systems.
Ok, to this scene enters SaaS solutions.
Most of them requires separate login credentials, their data are not accessible from other SaaS providers; their processes not accessible from your local processes etc... etc...
Does this not sound like the old legacy stovepipe solutions that we are trying so desperately to get away from ?
If we integrate SaaS solutions in the wrong way (or not at all) to our existing architecture we might just very soon find end users with 10 sticky notes containing passwords to 10 different SaaS applications complaining about doing swivel-chair integration of data between the systems.
Buying SaaS without strategic thinking; is there not a risk of taking a 5 year step back in creating an agile and flexible enterprise architecture ? Hmm....
This week I had an interesting afternoon with a really large organization and its strategic architects. They are in thinking mode for implementing a SOA strategy. We were discussing several things; one was
How-to convince the top management ?
It is a normal reaction not to react on future problems that are unlikely to impact you in the very near future(actions on climate change, pension savings etc..). People tend to react when they really, really need to - instead of having a long term strategy. Organizations are managed by people so the same rules more or less apply.
Today an organization really has 2 choices regarding its future IT architecture.
1. To plan and implement a SOA strategy
2. Not to plan and implement a SOA strategy
Now; if you bet on number 2 what will be the implications ? Well what will probably happen is that when new capabilities are requested the complexity of your future architecture will be much higher with to many integration points and unnecessary technology solutions. The cost to implement changes will continue to increment and one day you will find yourself in a situation where the cost of change is really out of proportion and then(!) you might reflect on number 1.
Many people are afraid to implement a SOA strategy today because they think that it will be very disruptive for the organization for many years, and yes it will be if you focus to much time and resources on it.
SOA is really about your long term strategy and being consistent ; your short term strategy can remain what you want it to be in the main time.
I was cited by IDC yesterday regarding standardized technology platforms.
Yes, there are some truth to that. The shift to standardize on communication protocols (WS-* etc...); which by the way gets more important; has created a world were the technology choice in the "backbone" is of less importance. IT of tomorrow will be based on functionality, capability and value and less of previous technology choices..
It is a good world :-) .
Finally online after many years of pressure from my friends :-) I plan to write about IT in general and IT architecture in particular. What I see happen in the industry and general observations.....
