Infrastructure as a Service

just few hours back I wrote this article & already we have a best usecase where in IaaS providers will gain , http://edition.cnn.com/2008/POLITICS/09/30/congress.website/index.html
this is exactly the LIVE unpresedented, un anticipated user bombardment which could happen to any application available online/internet, I feel if they had a IaaS solution they would have easily scaled up instaed of taking the site down , so in this case both the parties are in loss i.e. the visitiors who should have been able to view the legislative document which they could not & the gov in itself was not able to server its own stakeholders (citizens) who have the rights towards the service offered.

Honestly even though I have mastered the programming side of the IT world I get sleepless when it comes to deployments & managing the applications designed, be it the 24x7 up time, faster response, or scalability issues it always feel being challenged by the unknowns of this ONLINE WORLD of inter connected applications. Good news though now that services like Hadoop, Amazon EC2, SQS, SimpleDB or evn FPS it has shifted the power back to the developers in terms of getting to design & deploy scalable & expandable infrastructure on demand which I feel will make my job much easier when it comes to deploy on EC2/S3 as a file system while run the applications as an AMI (amazon instance), essentially that is helping a common developer by abstracting the low/system level challenges into fragmented API's those are easily understandable & usable within shorter span of time & that too without having to be the master of all the low level transactions. even though amazon has been building these systems since last few years I think we are on verge of seeing an exponential growth for companies & end users (programmers) who can build & be successful with very high end solutions, & in case they even remotely FAIL (obviously for the solution use case they tackle) they can retry/realign their product without having any impact on the infrastructure investments (which are base don how much you use ) & is bare minimum even in terms of $100 & NOT Millions what we have seen getting evaporated because of above issues.

Only things that concerns me are that while accessing these on "demand business objects" off the cloud we end up paying PER FETCH that could mean that the objects are expensive till they are ALIVE, so I am giving it a try to understand how much of cost advantage it is to manage a BUSINESS OBJECT off & on cloud based platform. Though initial cost seems to be minuscule though over time the archived data would becomes more expensive (I guess) which could be a major issue for large enterprises who have to archive huge chunks of data for Sarbens Oxley or any other corporate obligations....