Nov 18

I just came across The Economist’s debate on cloud computing. Stephen Elop, President of Microsoft’s Business Division and Marc Benioff, CEO of Salesforce outline their conflicting views on the future development of cloud computing. After a short scroll, I hold Benioff’s comments for clearly the more biased. I respect what Salesforce has achieved but Benioff’s views are still somewhat utopian. Elop on the other hand manages to deliver a much more realistic and less biased point of view. I am eager to see how Microsoft will realign or reshape their Azure strategy in general. I’m starting to think that some experts have prematurely underestimated Microsoft as the cloud losers.

All in all it’s great to follow these ongoing discussions. I’m especially interested in whether the outlooks I conceived in early 2009 (in my second bachelor’s thesis on cloud computing) will prove to be true in 1 or 2 years’ time. My personal opinion is that this technology is certainly not the “breakthrough” it is marketed to be. Rather, it’s a logical evolutionary step for virtualization combined with better utilization of existing computing capacities. Yet cloud services and cloud technology do make a lot of sense for a variety of business tasks and maybe even more technical tasks. But I guess I’m still skeptical because the cloud hype has to run its natural course first. Eventually, more mature solutions will settle in.

Tagged with:
Nov 15

Creative interview questions for consultancy companies and technology giants such as Google are nothing new. In fact the most commonly used of them are almost obsolete now and their “correct answers” are widespread. In fact one should definitely

google them

before going to Google’s interview. :) Questions such as How much should you charge to wash all the windows in Seattle? (obviously the market hourly rate or slightly above) or standard Fermi questions such as How many golf balls fit in a school bus? have a common solving pattern as far as a phone interview goes.

However, the popular question How long would it take to sort 1 trillion numbers? now has a very easy answer – 1 minute. The Hadoop project (led by Apache and Yahoo!) managed to benchmark a Petabyte sort in 16 hours, sorting a 1 Terabyte in 62 seconds. It seems that Yahoo! is very serious about bringing Hadoop to become the cloud computing platform. Yahoo’s strategy is to invest and lead the core developement of what is otherwise a free and open source project. They will eventually fully reap fruit and benefit once it really takes off and becomes the widespread standard for distributed applications/MapReduce framework. It will be interesting to follow how Google will actually respond and whether they would, at some later point in time share freely parts of their proprietary BigTable/MapReduce/GFS frameworks just to establish market share there too.

Tagged with:
preload preload preload