Sriram, Your comment on building services on webservers before going big bang using ESB or other integration tools will definitely simplify development.. In my experience Integrating these services costs more and in many ways require refactoring the existing components. Could you please share your view on ROI,
The approach works if your equation is something like this: Big upfront integration costs 100 units with a 50% chance of failure. Doing it simple first costs say 60 units with a 20% chance of failure. Later integration costs say another 70 units with a 25% chance of failure. Say we use a metric called effective cost = cost/(chance of success). Then the first approach costs 200 and the second two-stage approach costs 168. Besides you would have already realised some return by having the simple service go live earlier.
2 comments:
Sriram,
Your comment on building services on webservers before going big bang using ESB or other integration tools will definitely simplify development.. In my experience Integrating these services costs more and in many ways require refactoring the existing components.
Could you please share your view on ROI,
thanks,
vJ
vJ:
The approach works if your equation is something like this: Big upfront integration costs 100 units with a 50% chance of failure. Doing it simple first costs say 60 units with a 20% chance of failure. Later integration costs say another 70 units with a 25% chance of failure. Say we use a metric called effective cost = cost/(chance of success). Then the first approach costs 200 and the second two-stage approach costs 168. Besides you would have already realised some return by having the simple service go live earlier.
Post a Comment
Note: Only a member of this blog may post a comment.